<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="klubicka-kelleher-2023-probing">
    <titleInfo>
        <title>Probing Taxonomic and Thematic Embeddings for Taxonomic Information</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Filip</namePart>
        <namePart type="family">Klubička</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">John</namePart>
        <namePart type="family">Kelleher</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2023-01</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 12th Global Wordnet Conference</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">German</namePart>
            <namePart type="family">Rigau</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Francis</namePart>
            <namePart type="family">Bond</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Alexandre</namePart>
            <namePart type="family">Rademaker</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Global Wordnet Association</publisher>
            <place>
                <placeTerm type="text">University of the Basque Country, Donostia - San Sebastian, Basque Country</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Modelling taxonomic and thematic relatedness is important for building AI with comprehensive natural language understanding. The goal of this paper is to learn more about how taxonomic information is structurally encoded in embeddings. To do this, we design a new hypernym-hyponym probing task and perform a comparative probing study of taxonomic and thematic SGNS and GloVe embeddings. Our experiments indicate that both types of embeddings encode some taxonomic information, but the amount, as well as the geometric properties of the encodings, are independently related to both the encoder architecture, as well as the embedding training data. Specifically, we find that only taxonomic embeddings carry taxonomic information in their norm, which is determined by the underlying distribution in the data.</abstract>
    <identifier type="citekey">klubicka-kelleher-2023-probing</identifier>
    <location>
        <url>https://aclanthology.org/2023.gwc-1.1</url>
    </location>
    <part>
        <date>2023-01</date>
        <extent unit="page">
            <start>1</start>
            <end>13</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Probing Taxonomic and Thematic Embeddings for Taxonomic Information
%A Klubička, Filip
%A Kelleher, John
%Y Rigau, German
%Y Bond, Francis
%Y Rademaker, Alexandre
%S Proceedings of the 12th Global Wordnet Conference
%D 2023
%8 January
%I Global Wordnet Association
%C University of the Basque Country, Donostia - San Sebastian, Basque Country
%F klubicka-kelleher-2023-probing
%X Modelling taxonomic and thematic relatedness is important for building AI with comprehensive natural language understanding. The goal of this paper is to learn more about how taxonomic information is structurally encoded in embeddings. To do this, we design a new hypernym-hyponym probing task and perform a comparative probing study of taxonomic and thematic SGNS and GloVe embeddings. Our experiments indicate that both types of embeddings encode some taxonomic information, but the amount, as well as the geometric properties of the encodings, are independently related to both the encoder architecture, as well as the embedding training data. Specifically, we find that only taxonomic embeddings carry taxonomic information in their norm, which is determined by the underlying distribution in the data.
%U https://aclanthology.org/2023.gwc-1.1
%P 1-13

Markdown (Informal)

[Probing Taxonomic and Thematic Embeddings for Taxonomic Information](https://aclanthology.org/2023.gwc-1.1) (Klubička & Kelleher, GWC 2023)

• Probing Taxonomic and Thematic Embeddings for Taxonomic Information (Klubička & Kelleher, GWC 2023)

ACL

• Filip Klubička and John Kelleher. 2023. Probing Taxonomic and Thematic Embeddings for Taxonomic Information. In Proceedings of the 12th Global Wordnet Conference, pages 1–13, University of the Basque Country, Donostia - San Sebastian, Basque Country. Global Wordnet Association.