Subword regularization, used widely in NLP, improves model performance by reducing the dependency on exact tokenizations, augmenting the training corpus, and exposing the model to more unique contexts during training. BPE and MaxMatch, two popular subword tokenization schemes, have stochastic dropout regularization variants. However, there has not been an analysis of the distributions formed by them.We show that these stochastic variants are heavily biased towards a small set of tokenizations per word. If the benefits of subword regularization are as mentioned, we hypothesize that biasedness artificially limits the effectiveness of these schemes. Thus, we propose an algorithm to uniformly sample tokenizations that we use as a drop-in replacement for the stochastic aspects of existing tokenizers, and find that it improves machine translation quality.

We explore threshold vocabulary trimming in Byte-Pair Encoding subword tokenization, a tokenization postprocessing step that replaces rare subwords with their component subwords. The technique is available in popular tokenization libraries but has not been subjected to rigorous scientific scrutiny. While the removal of rare subwords is suggested as best practice in model implementations, both as a means to reduce model size and for improving model performance through robustness, our experiments indicate that, across a large space of hyperparameter settings, vocabulary trimming fails to consistently improve model performance, and is even prone to incurring heavy degradation.

Humans work together to solve common problems by having discussions, explaining, and agreeing or disagreeing with each other.Similarly, if a system can have discussions with human partners when solving tasks, it has the potential to improve the system’s performance and reliability.In previous research on explainability, it has only been possible for systems to make predictions and for humans to ask questions about them, rather than having a mutual exchange of opinions.This research aims to create a dataset and computational framework for systems that discuss and refine their predictions through dialogue. Through experiments, we show that the proposed system can have beneficial discussions with humans, improving the accuracy by up to 25 points on a natural language inference task.

Large Language Models (LLMs) are widely used to evaluate natural language generation tasks as automated metrics.However, the likelihood, a measure of LLM’s plausibility for a sentence, can vary due to superficial differences in sentences, such as word order and sentence structure.It is therefore possible that there might be a likelihood bias if LLMs are used for evaluation: they might overrate sentences with higher likelihoods while underrating those with lower likelihoods.In this paper, we investigate the presence and impact of likelihood bias in LLM-based evaluators.We also propose a method to mitigate the likelihood bias.Our method utilizes highly biased instances as few-shot examples for in-context learning.Our experiments in evaluating the data-to-text and grammatical error correction tasks reveal that several LLMs we test display a likelihood bias.Furthermore, our proposed method successfully mitigates this bias, also improving evaluation performance (in terms of correlation of models with human scores) significantly.

To combat the misuse of Large Language Models (LLMs), many recent studies have presented LLM-generated-text detectors with promising performance. When users instruct LLMs to generate texts, the instruction can include different constraints depending on the user’s need. However, most recent studies do not cover such diverse instruction patterns when creating datasets for LLM detection. In this paper, we reveal that even task-oriented constraints — constraints that would naturally be included in an instruction and are not related to detection-evasion — cause existing powerful detectors to have a large variance in detection performance. We focus on student essay writing as a realistic domain and manually create task-oriented constraints based on several factors for essay quality. Our experiments show that the standard deviation (SD) of current detector performance on texts generated by an instruction with such a constraint is significantly larger (up to an SD of 14.4 F1-score) than that by generating texts multiple times or paraphrasing the instruction. We also observe an overall trend where the constraints can make LLM detection more challenging than without them. Finally, our analysis indicates that the high instruction-following ability of LLMs fosters the large impact of such constraints on detection performance.

M2 and its variants are the most widely used automatic evaluation metrics for grammatical error correction (GEC), which calculate an F-score using a phrase-based alignment between sentences. However, it is not straightforward at all to align learner sentences containing errors to their correct sentences. In addition, alignment calculations are computationally expensive. We propose GREEN, an alignment-free F-score for GEC evaluation. GREEN treats a sentence as a multiset of n-grams and extracts edits between sentences by set operations instead of computing an alignment. Our experiments confirm that GREEN performs better than existing methods for the corpus-level metrics and comparably for the sentence-level metrics even without computing an alignment. GREEN is available at https://github.com/shotakoyama/green.

Document-level Relation Extraction (DocRE) is the task of extracting all semantic relationships from a document. While studies have been conducted on English DocRE, limited attention has been given to DocRE in non-English languages. This work delves into effectively utilizing existing English resources to promote DocRE studies in non-English languages, with Japanese as the representative case. As an initial attempt, we construct a dataset by transferring an English dataset to Japanese. However, models trained on such a dataset are observed to suffer from low recalls. We investigate the error cases and attribute the failure to different surface structures and semantics of documents translated from English and those written by native speakers. We thus switch to explore if the transferred dataset can assist human annotation on Japanese documents. In our proposal, annotators edit relation predictions from a model trained on the transferred dataset. Quantitative analysis shows that relation recommendations suggested by the model help reduce approximately 50% of the human edit steps compared with the previous approach. Experiments quantify the performance of existing DocRE models on our collected dataset, portraying the challenges of Japanese and cross-lingual DocRE.

In Grammatical Error Correction (GEC), it is crucial to ensure the user’s comprehension of a reason for correction. Existing studies present tokens, examples, and hints for corrections, but do not directly explain the reasons in natural language. Although methods that use Large Language Models (LLMs) to provide direct explanations in natural language have been proposed for various tasks, no such method exists for GEC. Generating explanations for GEC corrections involves aligning input and output tokens, identifying correction points, and presenting corresponding explanations consistently. However, it is not straightforward to specify a complex format to generate explanations, because explicit control of generation is difficult with prompts. This study introduces a method called controlled generation with Prompt Insertion (PI) so that LLMs can explain the reasons for corrections in natural language. In PI, LLMs first correct the input text, and then we automatically extract the correction points based on the rules. The extracted correction points are sequentially inserted into the LLM’s explanation output as prompts, guiding the LLMs to generate explanations for the correction points. We also create an Explainable GEC (XGEC) dataset of correction reasons by annotating NUCLE, CoNLL2013, and CoNLL2014. Although generations from GPT-3.5 and ChatGPT using original prompts miss some correction points, the generation control using PI can explicitly guide to describe explanations for all correction points, contributing to improved performance in generating correction reasons.

Discriminatory gender biases have been found in Pre-trained Language Models (PLMs) for multiple languages. In Natural Language Inference (NLI), existing bias evaluation methods have focused on the prediction results of one specific label out of three labels, such as neutral. However, such evaluation methods can be inaccurate since unique biased inferences are associated with unique prediction labels. Addressing this limitation, we propose a bias evaluation method for PLMs, called NLI-CoAL, which considers all the three labels of NLI task. First, we create three evaluation data groups that represent different types of biases. Then, we define a bias measure based on the corresponding label output of each data group. In the experiments, we introduce a meta-evaluation technique for NLI bias measures and use it to confirm that our bias measure can distinguish biased, incorrect inferences from non-biased incorrect inferences better than the baseline, resulting in a more accurate bias evaluation. We create the datasets in English, Japanese, and Chinese, and successfully validate the compatibility of our bias measure across multiple languages. Lastly, we observe the bias tendencies in PLMs of different languages. To our knowledge, we are the first to construct evaluation datasets and measure PLMs’ bias from NLI in Japanese and Chinese.

In Tokenization and the Noiseless Channel (Zouhar et al., 2023), Rényi efficiency is suggested as an intrinsic mechanism for evaluating a tokenizer: for NLP tasks, the tokenizer which leads to the highest Rényi efficiency of the unigram distribution should be chosen. The Rényi efficiency is thus treated as a predictor of downstream performance (e.g., predicting BLEU for a machine translation task), without the expensive step of training multiple models with different tokenizers. Although useful, the predictive power of this metric is not perfect, and the authors note there are additional qualities of a good tokenization scheme that Rényi efficiency alone cannot capture. We describe two variants of BPE tokenization which can arbitrarily increase Rényi efficiency while decreasing the downstream model performance. These counterexamples expose cases where Rényi efficiency fails as an intrinsic tokenization metric and thus give insight for building more accurate predictors.

We present a summary of the efforts to improve conference peer review that were implemented at ACL’23. This includes work with the goal of improving review quality, clearer workflow and decision support for the area chairs, as well as our efforts to improve paper-reviewer matching for various kinds of non- mainstream NLP work, and improve the overall incentives for all participants of the peer review process. We present analysis of the factors affecting peer review, identify the most problematic issues that the authors complained about, and provide suggestions for the future chairs. We hope that publishing such reports would (a) improve transparency in decision-making, (b) help the people new to the field to understand how the *ACL conferences work, (c) provide useful data for the future chairs and workshop organizers, and also academic work on peer review, and (d) provide useful context for the final program, as a source of information for meta-research on the structure and trajectory of the field of NLP.

Recent studies have suggested that neural language models learn and store a large amount of facts and commonsense knowledge from training data. The ability of language models to restore such knowledge is often evaluated via zero-shot cloze-style QA tasks. However, such evaluations rely only on prediction accuracy without punishing the systems for their mistakes, e.g., simply guessing or hallucinating likely answers. Selective prediction is a more informative evaluation framework that takes the confidence of predictions into account. Under the selective prediction setting, a model is evaluated not only by the number of correct predictions, but also by the ability to filter out dubious predictions by estimating the confidence of individual predictions. Such confidence-aware evaluation is crucial for determining whether to trust zero-shot predictions of language models. In this paper, we apply the selective prediction setting to an existing benchmark, LAMA probe, and conduct extensive experiments with recent neural language models and different confidence functions. We empirically show that our Selective-LAMA evaluation is more robust to the effect of simple guesses than the conventional accuracy-based evaluation. Our evaluation reveals the importance of the choice of confidence functions by showing that simply relying on token probabilities is not always the best choice. Further analysis shows that various confidence functions exhibit different preferences over predicted tokens for a given context.

A 360-degree image captures the entire scene without the limitations of a camera’s field of view, which makes it difficult to describe all the contexts in a single caption. We propose a novel task called Query-based Image Captioning (QuIC) for 360-degree images, where a query (words or short phrases) specifies the context to describe. This task is more challenging than the conventional image captioning task, which describes salient objects in images, as it requires fine-grained scene understanding to select the contents consistent with user’s intent based on the query. We construct a dataset for the new task that comprises 3,940 360-degree images and 18,459 pairs of queries and captions annotated manually. Experiments demonstrate that fine-tuning image captioning models further on our dataset can generate more diverse and controllable captions from multiple contexts of 360-degree images.

Generalization in Visual Question Answering (VQA) requires models to answer questions about images with contexts beyond the training distribution. Existing attempts primarily refine unimodal aspects, overlooking enhancements in multimodal aspects. Besides, diverse interpretations of the input lead to various modes of answer generation, highlighting the role of causal reasoning between interpreting and answering steps in VQA. Through this lens, we propose Cognitive pathways VQA (CopVQA) improving the multimodal predictions by emphasizing causal reasoning factors. CopVQA first operates a pool of pathways that capture diverse causal reasoning flows through interpreting and answering stages. Mirroring human cognition, we decompose the responsibility of each stage into distinct experts and a cognition-enabled component (CC). The two CCs strategically execute one expert for each stage at a time. Finally, we prioritize answer predictions governed by pathways involving both CCs while disregarding answers produced by either CC, thereby emphasizing causal reasoning and supporting generalization. Our experiments on real-life and medical data consistently verify that CopVQA improves VQA performance and generalization across baselines and domains. Notably, CopVQA achieves a new state-of-the-art (SOTA) on the PathVQA dataset and comparable accuracy to the current SOTA on VQA-CPv2, VQAv2, and VQA- RAD, with one-fourth of the model size.

Large Language Models (LLMs) have demonstrated remarkable performance in various tasks and gained significant attention. LLMs are also used for local sequence transduction tasks, including grammatical error correction (GEC) and formality style transfer, where most tokens in a source text are kept unchanged. However, the models that generate all target tokens in such tasks have a tendency to simply copy the input text as is, without making needed changes, because the difference between input and output texts is minimal in the training data. This is also inefficient because the computational cost grows quadratically with the target sequence length with Transformer. This paper proposes predicting edit spans for the source text for local sequence transduction tasks. Representing an edit span with a position of the source text and corrected tokens, we can reduce the length of the target sequence and the computational cost for inference. We apply instruction tuning for LLMs on the supervision data of edit spans. Experiments show that the proposed method achieves comparable performance to the baseline in four tasks, paraphrasing, formality style transfer, GEC, and text simplification, despite reducing the length of the target text by as small as 21%. Furthermore, we report that the task-specific fine-tuning with the proposed method achieved state-of-the-art performance in the four tasks.

Document-level relation extraction (DocRE) is the task of identifying all relations between each entity pair in a document. Evidence, defined as sentences containing clues for the relationship between an entity pair, has been shown to help DocRE systems focus on relevant texts, thus improving relation extraction. However, evidence retrieval (ER) in DocRE faces two major issues: high memory consumption and limited availability of annotations. This work aims at addressing these issues to improve the usage of ER in DocRE. First, we propose DREEAM, a memory-efficient approach that adopts evidence information as the supervisory signal, thereby guiding the attention modules of the DocRE system to assign high weights to evidence. Second, we propose a self-training strategy for DREEAM to learn ER from automatically-generated evidence on massive data without evidence annotations. Experimental results reveal that our approach exhibits state-of-the-art performance on the DocRED benchmark for both DocRE and ER. To the best of our knowledge, DREEAM is the first approach to employ ER self-training.

Character-level language modeling has been shown empirically to perform well on highly agglutinative or morphologically rich languages while using only a small fraction of the parameters required by (sub)word models. Korean fits nicely into this framework, except that, like other CJK languages, it has a very large character vocabulary of 11,172 unique syllables. However, unlike Japanese Kanji and Chinese Hanzi, each Korean syllable can be uniquely factored into a small set of subcharacters, called jamo. We explore a “three-hot” scheme, where we exploit the decomposability of Korean characters to model at the syllable level but using only jamo-level representations. We find that our three-hot embedding and decoding scheme alleviates the two major issues with prior syllable- and jamo-level models. Namely, it requires fewer than 1% of the embedding parameters of a syllable model, and it does not require tripling the sequence length, as with jamo models. In addition, it addresses a theoretical flaw in a prior three-hot modeling scheme. Our experiments show that, even when reducing the number of embedding parameters by 99.6% (from 11.4M to just 36k), our model suffers no loss in translation quality compared to the baseline syllable model.

Numerous types of social biases have been identified in pre-trained language models (PLMs), and various intrinsic bias evaluation measures have been proposed for quantifying those social biases. Prior works have relied on human annotated examples to compare existing intrinsic bias evaluation measures. However, this approach is not easily adaptable to different languages nor amenable to large scale evaluations due to the costs and difficulties when recruiting human annotators. To overcome this limitation, we propose a method to compare intrinsic gender bias evaluation measures without relying on human-annotated examples. Specifically, we create multiple bias-controlled versions of PLMs using varying amounts of male vs. female gendered sentences, mined automatically from an unannotated corpus using gender-related word lists. Next, each bias-controlled PLM is evaluated using an intrinsic bias evaluation measure, and the rank correlation between the computed bias scores and the gender proportions used to fine-tune the PLMs is computed. Experiments on multiple corpora and PLMs repeatedly show that the correlations reported by our proposed method that does not require human annotated examples are comparable to those computed using human annotated examples in prior work.

A promising approach for knowledge-based Word Sense Disambiguation (WSD) is to select the sense whose contextualized embeddings computed for its definition sentence are closest to those computed for a target word in a given sentence. This approach relies on the similarity of the sense and context embeddings computed by a pre-trained language model. We propose a semantic specialization for WSD where contextualized embeddings are adapted to the WSD task using solely lexical knowledge. The key idea is, for a given sense, to bring semantically related senses and contexts closer and send different/unrelated senses farther away. We realize this idea as the joint optimization of the Attract-Repel objective for sense pairs and the self-training objective for context-sense pairs while controlling deviations from the original embeddings. The proposed method outperformed previous studies that adapt contextualized embeddings. It achieved state-of-the-art performance on knowledge-based WSD when combined with the reranking heuristic that uses the sense inventory. We found that the similarity characteristics of specialized embeddings conform to the key idea. We also found that the (dis)similarity of embeddings between the related/different/unrelated senses correlates well with the performance of WSD.

This paper describes the Tokyo Tech and AIST system in the GenChal 2022 shared task, which is the first shared task of feedback comment generation. We adopted five methods: data cleaning, fine-tuning pre-trained models, correcting errors in learners’ sentences, appending a correcting operation, and filtering out irrelevant outputs. Our system achieved F1 = 43.4 on the test dataset.

Aspect sentiment quad prediction (ASQP) analyzes the aspect terms, opinion terms, sentiment polarity, and aspect categories in a text. One challenge in this task is the scarcity of data owing to the high annotation cost. Data augmentation techniques are commonly used to address this issue. However, existing approaches simply rewrite texts in the training data, restricting the semantic diversity of the generated data and impairing the quality due to the inconsistency between text and quads. To address these limitations, we augment quads and train a quads-to-text model to generate corresponding texts. Furthermore, we designed novel strategies to filter out low-quality data and balance the sample difficulty distribution of the augmented dataset. Empirical studies on two ASQP datasets demonstrate that our method outperforms other data augmentation methods and achieves state-of-the-art performance on the benchmarks. The source code will be released upon acceptance.

Large-scale pre-trained language models such as GPT-3 have shown remarkable performance across various natural language processing tasks. However, applying prompt-based methods with GPT-3 for Grammatical Error Correction (GEC) tasks and their controllability remains underexplored. Controllability in GEC is crucial for real-world applications, particularly in educational settings, where the ability to tailor feedback according to learner levels and specific error types can significantly enhance the learning process. This paper investigates the performance and controllability of prompt-based methods with GPT-3 for GEC tasks using zero-shot and few-shot setting. We explore the impact of task instructions and examples on GPT-3’s output, focusing on controlling aspects such as minimal edits, fluency edits, and learner levels. Our findings demonstrate that GPT-3 could effectively perform GEC tasks, outperforming existing supervised and unsupervised approaches. We also showed that GPT-3 could achieve controllability when appropriate task instructions and examples are given.

Masked Language Models (MLMs) pre-trained by predicting masked tokens on large corpora have been used successfully in natural language processing tasks for a variety of languages. Unfortunately, it was reported that MLMs also learn discriminative biases regarding attributes such as gender and race. Because most studies have focused on MLMs in English, the bias of MLMs in other languages has rarely been investigated. Manual annotation of evaluation data for languages other than English has been challenging due to the cost and difficulty in recruiting annotators. Moreover, the existing bias evaluation methods require the stereotypical sentence pairs consisting of the same context with attribute words (e.g. He/She is a nurse).We propose Multilingual Bias Evaluation (MBE) score, to evaluate bias in various languages using only English attribute word lists and parallel corpora between the target language and English without requiring manually annotated data. We evaluated MLMs in eight languages using the MBE and confirmed that gender-related biases are encoded in MLMs for all those languages. We manually created datasets for gender bias in Japanese and Russian to evaluate the validity of the MBE.The results show that the bias scores reported by the MBE significantly correlates with that computed from the above manually created datasets and the existing English datasets for gender bias.

Neural models trained with large amount of parallel data have achieved impressive performance in abstractive summarization tasks. However, large-scale parallel corpora are expensive and challenging to construct. In this work, we introduce a low-cost and effective strategy, ExtraPhrase, to augment training data for abstractive summarization tasks. ExtraPhrase constructs pseudo training data in two steps: extractive summarization and paraphrasing. We extract major parts of an input text in the extractive summarization step and obtain its diverse expressions with the paraphrasing step. Through experiments, we show that ExtraPhrase improves the performance of abstractive summarization tasks by more than 0.50 points in ROUGE scores compared to the setting without data augmentation. ExtraPhrase also outperforms existing methods such as back-translation and self-training. We also show that ExtraPhrase is significantly effective when the amount of genuine training data is remarkably small, i.e., a low-resource setting. Moreover, ExtraPhrase is more cost-efficient than the existing approaches

Formality style transfer (FST) is a task that involves paraphrasing an informal sentence into a formal one without altering its meaning. To address the data-scarcity problem of existing parallel datasets, previous studies tend to adopt a cycle-reconstruction scheme to utilize additional unlabeled data, where the FST model mainly benefits from target-side unlabeled sentences. In this work, we propose a simple yet effective semi-supervised framework to better utilize source-side unlabeled sentences based on consistency training. Specifically, our approach augments pseudo-parallel data obtained from a source-side informal sentence by enforcing the model to generate similar outputs for its perturbed version. Moreover, we empirically examined the effects of various data perturbation methods and propose effective data filtering strategies to improve our framework. Experimental results on the GYAFC benchmark demonstrate that our approach can achieve state-of-the-art results, even with less than 40% of the parallel data.

Grammatical Error Correction (GEC) should not focus only on high accuracy of corrections but also on interpretability for language learning. However, existing neural-based GEC models mainly aim at improving accuracy, and their interpretability has not been explored.A promising approach for improving interpretability is an example-based method, which uses similar retrieved examples to generate corrections. In addition, examples are beneficial in language learning, helping learners understand the basis of grammatically incorrect/correct texts and improve their confidence in writing. Therefore, we hypothesize that incorporating an example-based method into GEC can improve interpretability as well as support language learners. In this study, we introduce an Example-Based GEC (EB-GEC) that presents examples to language learners as a basis for a correction result. The examples consist of pairs of correct and incorrect sentences similar to a given input and its predicted correction. Experiments demonstrate that the examples presented by EB-GEC help language learners decide to accept or refuse suggestions from the GEC output. Furthermore, the experiments also show that retrieved examples improve the accuracy of corrections.

Logical table-to-text generation is a task that involves generating logically faithful sentences from tables, which requires models to derive logical-level facts from table records via logical inference. It raises a new challenge on the logical-level content planning of table-to-text models. However, directly learning the logical inference knowledge from table-text pairs is very difficult for neural models because of the ambiguity of natural language and the scarcity of parallel data. Hence even large-scale pre-trained language models present low logical fidelity on logical table-to-text. In this work, we propose a Pretrained Logical Form Generator (PLOG) framework to improve generation fidelity. Specifically, PLOG is first pretrained on a table-to-logical-form generation (table-to-logic) task, then finetuned on downstream table-to-text tasks. The logical forms are formally defined with unambiguous semantics. Hence we can collect a large amount of accurate logical forms from tables without human annotation. In addition, PLOG can learn logical inference from table-logic pairs much more reliably than from table-text pairs. To evaluate our model, we further collect a controlled logical table-to-text dataset CONTLOG based on an existing dataset. On two benchmarks, LOGICNLG and CONTLOG, PLOG outperforms strong baselines by a large margin on the logical fidelity, demonstrating the effectiveness of table-to-logic pretraining.

Subword regularizations use multiple subword segmentations during training to improve the robustness of neural machine translation models. In previous subword regularizations, we use multiple segmentations in the training process but use only one segmentation in the inference. In this study, we propose an inference strategy to address this discrepancy. The proposed strategy approximates the marginalized likelihood by using multiple segmentations including the most plausible segmentation and several sampled segmentations. Because the proposed strategy aggregates predictions from several segmentations, we can regard it as a single model ensemble that does not require any additional cost for training. Experimental results show that the proposed strategy improves the performance of models trained with subword regularization in low-resource machine translation tasks.

We present two simple modifications for word-level perturbation: Word Replacement considering Length (WR-L) and Compositional Word Replacement (CWR).In conventional word replacement, a word in an input is replaced with a word sampled from the entire vocabulary, regardless of the length and context of the target word.WR-L considers the length of a target word by sampling words from the Poisson distribution.CWR considers the compositional candidates by restricting the source of sampling to related words that appear in subword regularization. Experimental results showed that the combination of WR-L and CWR improved the performance of text classification and machine translation.

Different methods have been proposed to develop meta-embeddings from a given set of source embeddings. However, the source embeddings can contain unfair gender-related biases, and how these influence the meta-embeddings has not been studied yet.We study the gender bias in meta-embeddings created under three different settings:(1) meta-embedding multiple sources without performing any debiasing (Multi-Source No-Debiasing),(2) meta-embedding multiple sources debiased by a single method (Multi-Source Single-Debiasing), and(3) meta-embedding a single source debiased by different methods (Single-Source Multi-Debiasing).Our experimental results show that meta-embedding amplifies the gender biases compared to input source embeddings.We find that debiasing not only the sources but also their meta-embedding is needed to mitigate those biases.Moreover, we propose a novel debiasing method based on meta-embedding learning where we use multiple debiasing methods on a single source embedding and then create a single unbiased meta-embedding.

This study introduces a novel approach to the joint extraction of entities and relations by stacking convolutional neural networks (CNNs) on pretrained language models. We adopt table representations to model the entities and relations, casting the entity and relation extraction as a table-labeling problem. Regarding each table as an image and each cell in a table as an image pixel, we apply two-dimensional CNNs to the tables to capture local dependencies and predict the cell labels. The experimental results showed that the performance of the proposed method is comparable to those of current state-of-art systems on the CoNLL04, ACE05, and ADE datasets. Even when freezing pretrained language model parameters, the proposed method showed a stable performance, whereas the compared methods suffered from significant decreases in performance. This observation indicates that the parameters of the pretrained encoder may incorporate dependencies among the entity and relation labels during fine-tuning.

We present a multi-task learning framework for cross-lingual abstractive summarization to augment training data. Recent studies constructed pseudo cross-lingual abstractive summarization data to train their neural encoder-decoders. Meanwhile, we introduce existing genuine data such as translation pairs and monolingual abstractive summarization data into training. Our proposed method, Transum, attaches a special token to the beginning of the input sentence to indicate the target task. The special token enables us to incorporate the genuine data into the training data easily. The experimental results show that Transum achieves better performance than the model trained with only pseudo cross-lingual summarization data. In addition, we achieve the top ROUGE score on Chinese-English and Arabic-English abstractive summarization. Moreover, Transum also has a positive effect on machine translation. Experimental results indicate that Transum improves the performance from the strong baseline, Transformer, in Chinese-English, Arabic-English, and English-Japanese translation datasets.

Korean is a language with complex morphology that uses spaces at larger-than-word boundaries, unlike other East-Asian languages. While morpheme-based text generation can provide significant semantic advantages compared to commonly used character-level approaches, Korean morphological analyzers only provide a sequence of morpheme-level tokens, losing information in the tokenization process. Two crucial issues are the loss of spacing information and subcharacter level morpheme normalization, both of which make the tokenization result challenging to reconstruct the original input string, deterring the application to generative tasks. As this problem originates from the conventional scheme used when creating a POS tagging corpus, we propose an improvement to the existing scheme, which makes it friendlier to generative tasks. On top of that, we suggest a fully-automatic annotation of a corpus by leveraging public analyzers. We vote the surface and POS from the outcome and fill the sequence with the selected morphemes, yielding tokenization with a decent quality that incorporates space information. Our scheme is verified via an evaluation done on an external corpus, and subsequently, it is adapted to Korean Wikipedia to construct an open, permissive resource. We compare morphological analyzer performance trained on our corpus with existing methods, then perform an extrinsic evaluation on a downstream task.

South and North Korea both use the Korean language. However, Korean NLP research has focused on South Korean only, and existing NLP systems of the Korean language, such as neural machine translation (NMT) models, cannot properly handle North Korean inputs. Training a model using North Korean data is the most straightforward approach to solving this problem, but there is insufficient data to train NMT models. In this study, we create data for North Korean NMT models using a comparable corpus. First, we manually create evaluation data for automatic alignment and machine translation, and then, investigate automatic alignment methods suitable for North Korean. Finally, we show that a model trained by North Korean bilingual data without human annotation significantly boosts North Korean translation accuracy compared to existing South Korean models in zero-shot settings.

IR models using a pretrained language model significantly outperform lexical approaches like BM25. In particular, SPLADE, which encodes texts to sparse vectors, is an effective model for practical use because it shows robustness to out-of-domain datasets. However, SPLADE still struggles with exact matching of low-frequency words in training data. In addition, domain shifts in vocabulary and word frequencies deteriorate the IR performance of SPLADE. Because supervision data are scarce in the target domain, addressing the domain shifts without supervision data is necessary. This paper proposes an unsupervised domain adaptation method by filling vocabulary and word-frequency gaps. First, we expand a vocabulary and execute continual pretraining with a masked language model on a corpus of the target domain. Then, we multiply SPLADE-encoded sparse vectors by inverse document frequency weights to consider the importance of documents with low-frequency words. We conducted experiments using our method on datasets with a large vocabulary gap from a source domain. We show that our method outperforms the present state-of-the-art domain adaptation method. In addition, our method achieves state-of-the-art results, combined with BM25.

We study the relationship between task-agnostic intrinsic and task-specific extrinsic social bias evaluation measures for MLMs, and find that there exists only a weak correlation between these two types of evaluation measures. Moreover, we find that MLMs debiased using different methods still re-learn social biases during fine-tuning on downstream tasks. We identify the social biases in both training instances as well as their assigned labels as reasons for the discrepancy between intrinsic and extrinsic bias evaluation measurements. Overall, our findings highlight the limitations of existing MLM bias evaluation measures and raise concerns on the deployment of MLMs in downstream applications using those measures.

Automatic evaluation of grammatical error correction (GEC) is essential in developing useful GEC systems. Existing methods for automatic evaluation require multiple reference sentences or manual scores. However, such resources are expensive, thereby hindering automatic evaluation for various domains and correction styles. This paper proposes an Impact-based Metric for GEC using PARAllel data, IMPARA, which utilizes correction impacts computed by parallel data comprising pairs of grammatical/ungrammatical sentences. As parallel data is cheaper than manually assessing evaluation scores, IMPARA can reduce the cost of data creation for automatic evaluation. Correlations between IMPARA and human scores indicate that IMPARA is comparable or better than existing evaluation methods. Furthermore, we find that IMPARA can perform evaluations that fit different domains and correction styles trained on various parallel data.

Large-scale pretraining and task-specific fine- tuning is now the standard methodology for many tasks in computer vision and natural language processing. Recently, a multitude of methods have been proposed for pretraining vision and language BERTs to tackle challenges at the intersection of these two key areas of AI. These models can be categorized into either single-stream or dual-stream encoders. We study the differences between these two categories, and show how they can be unified under a single theoretical framework. We then conduct controlled experiments to discern the empirical differences between five vision and language BERTs. Our experiments show that training data and hyperparameters are responsible for most of the differences between the reported results, but they also reveal that the embedding layer plays a crucial role in these massive models.

Field experiments on a foreign news distribution system using two key technologies are reported. The first technology is a summarization component, which is used for generating news headlines. This component is a transformer-based abstractive text summarization system which is trained to output headlines from the leading sentences of news articles. The second technology is machine translation (MT), which enables users to read foreign news articles in their mother language. Since the system uses MT, users can immediately access the latest foreign news. 139 Japanese LINE users participated in the field experiments for two weeks, viewing about 40,000 articles which had been translated from English to Japanese. We carried out surveys both during and after the experiments. According to the results, 79.3% of users evaluated the headlines as adequate, while 74.7% of users evaluated the automatically translated articles as intelligible. According to the post-experiment survey, 59.7% of users wished to continue using the system; 11.5% of users did not. We also report several statistics of the experiments.

We address the task of antonym prediction in a context, which is a fill-in-the-blanks problem. This task setting is unique and practical because it requires contrastiveness to the other word and naturalness as a text in filling a blank. We propose methods for fine-tuning pre-trained masked language models (BERT) for context-aware antonym prediction. The experimental results demonstrate that these methods have positive impacts on the prediction of antonyms within a context. Moreover, human evaluation reveals that more than 85% of predictions using the proposed method are acceptable as antonyms.

This paper explores a variant of automatic headline generation methods, where a generated headline is required to include a given phrase such as a company or a product name. Previous methods using Transformer-based models generate a headline including a given phrase by providing the encoder with additional information corresponding to the given phrase. However, these methods cannot always include the phrase in the generated headline. Inspired by previous RNN-based methods generating token sequences in backward and forward directions from the given phrase, we propose a simple Transformer-based method that guarantees to include the given phrase in the high-quality generated headline. We also consider a new headline generation strategy that takes advantage of the controllable generation order of Transformer. Our experiments with the Japanese News Corpus demonstrate that our methods, which are guaranteed to include the phrase in the generated headline, achieve ROUGE scores comparable to previous Transformer-based methods. We also show that our generation strategy performs better than previous strategies.

In this paper, we address the task of news-image captioning, which generates a description of an image given the image and its article body as input. This task is more challenging than the conventional image captioning, because it requires a joint understanding of image and text. We present a Transformer model that integrates text and image modalities and attends to textual features from visual features in generating a caption. Experiments based on automatic evaluation metrics and human evaluation show that an article text provides primary information to reproduce news-image captions written by journalists. The results also demonstrate that the proposed model outperforms the state-of-the-art model. In addition, we also confirm that visual features contribute to improving the quality of news-image captions.

Explainable recommendation is a good way to improve user satisfaction. However, explainable recommendation in dialogue is challenging since it has to handle natural language as both input and output. To tackle the challenge, this paper proposes a novel and practical task to explain evidences in recommending hotels given vague requests expressed freely in natural language. We decompose the process into two subtasks on hotel reviews: Evidence Identification and Evidence Explanation. The former predicts whether or not a sentence contains evidence that expresses why a given request is satisfied. The latter generates a recommendation sentence given a request and an evidence sentence. In order to address these subtasks, we build an Evidence-based Explanation dataset, which is the largest dataset for explaining evidences in recommending hotels for vague requests. The experimental results demonstrate that the BERT model can find evidence sentences with respect to various vague requests and that the LSTM-based model can generate recommendation sentences.

We describe the system submitted by the SWAGex team to the SemEval-2020 Commonsense Validation and Explanation Task. We use multiple methods on the pre-trained language model BERT (Devlin et al., 2018) for tasks that require the system to recognize sentences against commonsense and justify the reasoning behind this decision. Our best performing model is BERT trained on SWAG and fine-tuned for the task. We investigate the ability to transfer commonsense knowledge from SWAG to SemEval-2020 by training a model for the Explanation task with Next Event Prediction data

This paper describes the emphasis selection system of the team TextLearner for SemEval 2020 Task 10: Emphasis Selection For Written Text in Visual Media. The system aims to learn the emphasis selection distribution using contextual representations extracted from pre-trained language models and a two-staged ranking model. The experimental results demonstrate the strong contextual representation power of the recent advanced transformer-based language model RoBERTa, which can be exploited using a simple but effective architecture on top.

In the context of multilingual language model pre-training, vocabulary size for languages with a broad set of potential characters is an unsolved problem. We propose two algorithms applicable in any unsupervised multilingual pre-training task, increasing the elasticity of budget required for building the vocabulary in Byte-Pair Encoding inspired tokenizers, significantly reducing the cost of supporting Korean in a multilingual model.

In natural language, we often omit some words that are easily understandable from the context. In particular, pronouns of subject, object, and possessive cases are often omitted in Japanese; these are known as zero pronouns. In translation from Japanese to other languages, we need to find a correct antecedent for each zero pronoun to generate a correct and coherent translation. However, it is difficult for conventional automatic evaluation metrics (e.g., BLEU) to focus on the success of zero pronoun resolution. Therefore, we present a hand-crafted dataset to evaluate whether translation models can resolve the zero pronoun problems in Japanese to English translations. We manually and statistically validate that our dataset can effectively evaluate the correctness of the antecedents selected in translations. Through the translation experiments using our dataset, we reveal shortcomings of an existing context-aware neural machine translation model.

Most studies on abstractive summarization report ROUGE scores between system and reference summaries. However, we have a concern about the truthfulness of generated summaries: whether all facts of a generated summary are mentioned in the source text. This paper explores improving the truthfulness in headline generation on two popular datasets. Analyzing headlines generated by the state-of-the-art encoder-decoder model, we show that the model sometimes generates untruthful headlines. We conjecture that one of the reasons lies in untruthful supervision data used for training the model. In order to quantify the truthfulness of article-headline pairs, we consider the textual entailment of whether an article entails its headline. After confirming quite a few untruthful instances in the datasets, this study hypothesizes that removing untruthful instances from the supervision data may remedy the problem of the untruthful behaviors of the model. Building a binary classifier that predicts an entailment relation between an article and its headline, we filter out untruthful instances from the supervision data. Experimental results demonstrate that the headline generation model trained on filtered supervision data shows no clear difference in ROUGE scores but remarkable improvements in automatic and manual evaluations of the generated headlines.

Most neural machine translation models only rely on pairs of parallel sentences, assuming syntactic information is automatically learned by an attention mechanism. In this work, we investigate different approaches to incorporate syntactic knowledge in the Transformer model and also propose a novel, parameter-free, dependency-aware self-attention mechanism that improves its translation quality, especially for long sentences and in low-resource scenarios. We show the efficacy of each approach on WMT English-German and English-Turkish, and WAT English-Japanese translation tasks.

The performance of neural machine translation systems is commonly evaluated in terms of BLEU. However, due to its reliance on target language properties and generation, the BLEU metric does not allow an assessment of which translation directions are more difficult to model. In this paper, we propose cross-mutual information (XMI): an asymmetric information-theoretic metric of machine translation difficulty that exploits the probabilistic nature of most neural machine translation models. XMI allows us to better evaluate the difficulty of translating text into the target language while controlling for the difficulty of the target-side generation component independent of the translation task. We then present the first systematic and controlled study of cross-lingual translation difficulties using modern neural translation systems. Code for replicating our experiments is available online at https://github.com/e-bug/nmt-difficulty.

In traditional NLP, we tokenize a given sentence as a preprocessing, and thus the tokenization is unrelated to a target downstream task. To address this issue, we propose a novel method to explore a tokenization which is appropriate for the downstream task. Our proposed method, optimizing tokenization (OpTok), is trained to assign a high probability to such appropriate tokenization based on the downstream task loss. OpTok can be used for any downstream task which uses a vector representation of a sentence such as text classification. Experimental results demonstrate that OpTok improves the performance of sentiment analysis and textual entailment. In addition, we introduce OpTok into BERT, the state-of-the-art contextualized embeddings and report a positive effect.

Large scale pre-trained language models have shown groundbreaking performance improvements for transfer learning in the domain of natural language processing. In our paper, we study a pre-trained multilingual BERT model and analyze the OOV rate on downstream tasks, how it introduces information loss, and as a side-effect, obstructs the potential of the underlying model. We then propose multiple approaches for mitigation and demonstrate that it improves performance with the same parameter count when combined with fine-tuning.

This paper presents our contextual emotion detection system in approaching the SemEval2019 shared task 3: EmoContext: Contextual Emotion Detection in Text. This system cooperates with an emotion detection neural network method (Poria et al., 2017), emoji2vec (Eisner et al., 2016) embedding, word2vec embedding (Mikolov et al., 2013), and our proposed emoticon and emoji preprocessing method. The experimental results demonstrate the usefulness of our emoticon and emoji prepossessing method, and representations of emoticons and emoji contribute model’s emotion detection.

We propose a data-to-text generation model with two modules, one for tracking and the other for text generation. Our tracking module selects and keeps track of salient information and memorizes which record has been mentioned. Our generation module generates a summary conditioned on the state of tracking module. Our proposed model is considered to simulate the human-like writing process that gradually selects the information by determining the intermediate variables while writing the summary. In addition, we also explore the effectiveness of the writer information for generations. Experimental results show that our proposed model outperforms existing models in all evaluation metrics even without writer information. Incorporating writer information further improves the performance, contributing to content planning and surface realization.

Neural encoder-decoder models have been successful in natural language generation tasks. However, real applications of abstractive summarization must consider an additional constraint that a generated summary should not exceed a desired length. In this paper, we propose a simple but effective extension of a sinusoidal positional encoding (Vaswani et al., 2017) so that a neural encoder-decoder model preserves the length constraint. Unlike previous studies that learn length embeddings, the proposed method can generate a text of any length even if the target length is unseen in training data. The experimental results show that the proposed method is able not only to control generation length but also improve ROUGE scores.

Question Generation (QG) is the task of generating questions from a given passage. One of the key requirements of QG is to generate a question such that it results in a target answer. Previous works used a target answer to obtain a desired question. However, we also want to specify how to ask questions and improve the quality of generated questions. In this study, we explore the use of interrogative phrases as additional sources to control QG. By providing interrogative phrases, we expect that QG can generate a more reliable sequence of words subsequent to an interrogative phrase. We present a baseline sequence-to-sequence model with the attention, copy, and coverage mechanisms, and show that the simple baseline achieves state-of-the-art performance. The experiments demonstrate that interrogative phrases contribute to improving the performance of QG. In addition, we report the superiority of using interrogative phrases in human evaluation. Finally, we show that a question answering system can provide target answers more correctly when the questions are generated with interrogative phrases.

Browsing news articles on multiple devices is now possible. The lengths of news article headlines have precise upper bounds, dictated by the size of the display of the relevant device or interface. Therefore, controlling the length of headlines is essential when applying the task of headline generation to news production. However, because there is no corpus of headlines of multiple lengths for a given article, previous research on controlling output length in headline generation has not discussed whether the system outputs could be adequately evaluated without multiple references of different lengths. In this paper, we introduce two corpora, which are Japanese News Corpus (JNC) and JApanese MUlti-Length Headline Corpus (JAMUL), to confirm the validity of previous evaluation settings. The JNC provides common supervision data for headline generation. The JAMUL is a large-scale evaluation dataset for headlines of three different lengths composed by professional editors. We report new findings on these corpora; for example, although the longest length reference summary can appropriately evaluate the existing methods controlling output length, this evaluation setting has several problems.

Social media provide platforms to express, discuss, and shape opinions about events and issues in the real world. An important step to analyze the discussions on social media and to assist in healthy decision-making is stance detection. This paper presents an approach to detect the stance of a user toward a topic based on their stances toward other topics and the social media posts of the user. We apply factorization machines, a widely used method in item recommendation, to model user preferences toward topics from the social media data. The experimental results demonstrate that users’ posts are useful to model topic preferences and therefore predict stances of silent users.

Developing a method for understanding the inner workings of black-box neural methods is an important research endeavor. Conventionally, many studies have used an attention matrix to interpret how Encoder-Decoder-based models translate a given source sentence to the corresponding target sentence. However, recent studies have empirically revealed that an attention matrix is not optimal for token-wise translation analyses. We propose a method that explicitly models the token-wise alignment between the source and target sequences to provide a better analysis. Experiments show that our method can acquire token-wise alignments that are superior to those of an attention mechanism.

Temporal reasoning remains as an unsolved task for Natural Language Processing (NLP), particularly demonstrated in the clinical domain. The complexity of temporal representation in language is evident as results of the 2016 Clinical TempEval challenge indicate: the current state-of-the-art systems perform well in solving mention-identification tasks of event and time expressions but poorly in temporal relation extraction, showing a gap of around 0.25 point below human performance. We explore to adapt the tree-based LSTM-RNN model proposed by Miwa and Bansal (2016) to temporal relation extraction from clinical text, obtaining a five point improvement over the best 2016 Clinical TempEval system and two points over the state-of-the-art. We deliver a deep analysis of the results and discuss the next step towards human-like temporal reasoning.

This study addresses the problem of identifying the meaning of unknown words or entities in a discourse with respect to the word embedding approaches used in neural language models. We proposed a method for on-the-fly construction and exploitation of word embeddings in both the input and output layers of a neural model by tracking contexts. This extends the dynamic entity representation used in Kobayashi et al. (2016) and incorporates a copy mechanism proposed independently by Gu et al. (2016) and Gulcehre et al. (2016). In addition, we construct a new task and dataset called Anonymized Language Modeling for evaluating the ability to capture word meanings while reading. Experiments conducted using our novel dataset show that the proposed variant of RNN language model outperformed the baseline model. Furthermore, the experiments also demonstrate that dynamic updates of an output layer help a model predict reappearing entities, whereas those of an input layer are effective to predict words following reappearing entities.

This paper explores the idea of robot editors, automated proofreaders that enable journalists to improve the quality of their articles. We propose a novel neural model of multi-task learning that both generates proofread sentences and predicts the editing operations required to rewrite the source sentences and create the proofread ones. The model is trained using logs of the revisions made professional editors revising draft newspaper articles written by journalists. Experiments demonstrate the effectiveness of our multi-task learning approach and the potential value of using revision logs for this task.

We address the issue of the quality of journalism and analyze daily article revision logs from a Japanese newspaper company. The revision logs contain data that can help reveal the requirements of quality journalism such as the types and number of edit operations and aspects commonly focused in revision. This study also discusses potential applications such as quality assessment and automatic article revision as our future research directions.

We presents in this paper our approach for modeling inter-topic preferences of Twitter users: for example, “those who agree with the Trans-Pacific Partnership (TPP) also agree with free trade”. This kind of knowledge is useful not only for stance detection across multiple topics but also for various real-world applications including public opinion survey, electoral prediction, electoral campaigns, and online debates. In order to extract users’ preferences on Twitter, we design linguistic patterns in which people agree and disagree about specific topics (e.g., “A is completely wrong”). By applying these linguistic patterns to a collection of tweets, we extract statements agreeing and disagreeing with various topics. Inspired by previous work on item recommendation, we formalize the task of modeling inter-topic preferences as matrix factorization: representing users’ preference as a user-topic matrix and mapping both users and topics onto a latent feature space that abstracts the preferences. Our experimental results demonstrate both that our presented approach is useful in predicting missing preferences of users and that the latent vector representations of topics successfully encode inter-topic preferences.

Automatic video description generation has recently been getting attention after rapid advancement in image caption generation. Automatically generating description for a video is more challenging than for an image due to its temporal dynamics of frames. Most of the work relied on Recurrent Neural Network (RNN) and recently attentional mechanisms have also been applied to make the model learn to focus on some frames of the video while generating each word in a describing sentence. In this paper, we focus on a sequence-to-sequence approach with temporal attention mechanism. We analyze and compare the results from different attention model configuration. By applying the temporal attention mechanism to the system, we can achieve a METEOR score of 0.310 on Microsoft Video Description dataset, which outperformed the state-of-the-art system so far.

This study proposes a computational model of the discourse segments in lyrics to understand and to model the structure of lyrics. To test our hypothesis that discourse segmentations in lyrics strongly correlate with repeated patterns, we conduct the first large-scale corpus study on discourse segments in lyrics. Next, we propose the task to automatically identify segment boundaries in lyrics and train a logistic regression model for the task with the repeated pattern and textual features. The results of our empirical experiments illustrate the significance of capturing repeated patterns in predicting the boundaries of discourse segments in lyrics.

This paper proposes a novel problem setting of selectional preference (SP) between a predicate and its arguments, called as context-sensitive SP (CSP). CSP models the narrative consistency between the predicate and preceding contexts of its arguments, in addition to the conventional SP based on semantic types. Furthermore, we present a novel CSP model that extends the neural SP model (Van de Cruys, 2014) to incorporate contextual information into the distributed representations of arguments. Experimental results demonstrate that the proposed CSP model successfully learns CSP and outperforms the conventional SP model in coreference cluster ranking.

We present the UOT Machine Translation System that was used in the IWSLT-09 evaluation campaign. This year, we participated in the BTEC track for Chinese-to-English translation. Our system is based on a string-to-tree framework. To integrate deep syntactic information, we propose the use of parse trees and semantic dependencies on English sentences described respectively by Head-driven Phrase Structure Grammar and Predicate-Argument Structures. We report the results of our system on both the development and test sets.

This paper proposes a method of increasing the size of a bilingual lexicon obtained from two other bilingual lexicons via a pivot language. When we apply this approach, there are two main challenges, “ambiguity” and “mismatch” of terms; we target the latter problem by improving the utilization ratio of the bilingual lexicons. Given two bilingual lexicons between language pairs Lf-Lp and Lp-Le, we compute lexical translation probabilities of word pairs by using a statistical word-alignment model, and term decomposition/composition techniques. We compare three approaches to generate the bilingual lexicon: “exact merging”, “word-based merging”, and our proposed “alignment-based merging”. In our method, we combine lexical translation probabilities and a simple language model for estimating the probabilities of translation pairs. The experimental results show that our method could drastically improve the number of translation terms compared to the two methods mentioned above. Additionally, we evaluated and discussed the quality of the translation outputs.

Many systems have been developed in the past few years to assist researchers in the discovery of knowledge published as English text, for example in the PubMed database. At the same time, higher level collective knowledge is often published using a graphical notation representing all the entities in a pathway and their interactions. We believe that these pathway visualizations could serve as an effective user interface for knowledge discovery if they can be linked to the text in publications. Since the graphical elements in a Pathway are of a very different nature than their corresponding descriptions in English text, we developed a prototype system called PathText. The goal of PathText is to serve as a bridge between these two different representations. In this paper, we first describe the overall architecture and the interfaces of the PathText system, and then provide some details about the core Text Mining components.

The continuous emergence of new technical terms and the difficulty of keeping up with neologism in parallel corpora deteriorate the performance of statistical machine translation (SMT) systems. This paper explores the use of morphological information to improve English-to-Chinese translation for technical terms. To reduce the morpheme-level translation ambiguity, we group the morphemes into morpheme phrases and propose the use of domain information for translation candidate selection. In order to find correspondences of morpheme phrases between the source and target languages, we propose an algorithm to mine morpheme phrase translation pairs from a bilingual lexicon. We also build a cascaded translation model that dynamically shifts translation units from phrase level to word and morpheme phrase levels. The experimental results show the significant improvements over the current phrase-based SMT systems.

Given the increasing number of neologisms in biomedicine (names of genes, diseases, molecules, etc.), the rate of acronyms used in literature also increases. Existing acronym dictionaries cannot keep up with the rate of new creations. Thus, discovering and disambiguating acronyms and their expanded forms are essential aspects of text mining and terminology management. We present a method for clustering long forms identified by an acronym recognition method. Applying the acronym recognition method to MEDLINE abstracts, we obtained a list of short/long forms. The recognized short/long forms were classified by abiologist to construct an evaluation set for clustering sets of similar long forms. We observed five types of term variation in the evaluation set and defined four similarity measures to gathers the similar longforms (i.e., orthographic, morphological, syntactic, lexico semantic variants, nested abbreviations). The complete-link clustering with the four similarity measures achieved 87.5% precision and 84.9% recall on the evaluation set.

One of the main challenges in biomedical text mining is the identification of terminology, which is a key factor for accessing and integrating the information stored in literature. Manual creation of biomedical terminologies cannot keep pace with the data that becomes available. Still, many of them have been used in attempts to recognise terms in literature, but their suitability for text mining has been questioned as substantial re-engineering is needed to tailor the resources for automatic processing. Several approaches have been suggested to automatically integrate and map between resources, but the problems of extensive variability of lexical representations and ambiguity have been revealed. In this paper we present a methodology to automatically maintain a biomedical terminological database, which contains automatically extracted terms, their mutual relationships, features and possible annotations that can be useful in text processing. In addition to TermDB, a database used for terminology management and storage, we present the following modules that are used to populate the database: TerMine (recognition, extraction and normalisation of terms from literature), AcroTerMine (extraction and clustering of acronyms and their long forms), AnnoTerm (annotation and classification of terms), and ClusTerm (extraction of term associations and clustering of terms).