\documentclass[11pt,round]{article}

% Change "review" to "final" to generate the final (sometimes called camera-ready) version.
% Change to "preprint" to generate a non-anonymous version with page numbers.
\usepackage[final]{acl}

% Standard package includes
\usepackage{times}
\usepackage{latexsym}
\usepackage{longtable} 
% For proper rendering and hyphenation of words containing Latin characters (including in bib files)
\usepackage[T1]{fontenc}
% For Vietnamese characters
% \usepackage[T5]{fontenc}
% See https://www.latex-project.org/help/documentation/encguide.pdf for other character sets

% This assumes your files are encoded as UTF8
\usepackage[utf8]{inputenc}

% This is not strictly necessary, and may be commented out,
% but it will improve the layout of the manuscript,
% and will typically save some space.
\usepackage{microtype}

% This is also not strictly necessary, and may be commented out.
% However, it will improve the aesthetics of text in
% the typewriter font.
\usepackage{inconsolata}

%Including images in your LaTeX document requires adding
%additional package(s)
\usepackage{graphicx}
%\usepackage[round]{natbib}
% If the title and author information does not fit in the area allocated, uncomment the following
%
%\setlength\titlebox{<dim>}
%
% and set <dim> to something 5cm or larger.

\title{Temporal and Geographic Patterns in Child Sexual Abuse Prosecutions: A Decade of Magistrates’ Court Records from England and Wales}

% Author information can be set in various styles:
% For several authors from the same institution:
% \author{Author 1 \and ... \and Author n \\
%         Address line \\ ... \\ Address line}
% if the names do not fit well on one line use
%         Author 1 \\ {\bf Author 2} \\ ... \\ {\bf Author n} \\
% For authors from different institutions:
% \author{Author 1 \\ Address line \\  ... \\ Address line
%         \And  ... \And
%         Author n \\ Address line \\ ... \\ Address line}
% To start a separate ``row'' of authors use \AND, as in
% \author{Author 1 \\ Address line \\  ... \\ Address line
%         \AND
%         Author 2 \\ Address line \\ ... \\ Address line \And
%         Author 3 \\ Address line \\ ... \\ Address line}

\author{Tabassom Sedighi \\
  Anglia Ruskin University \\
   Bishop Hall Ln, Chelmsford\\
  CM1 1SQ, United Kingdom \\
  \texttt{tabassom.sedighi@aru.ac.uk} \\\And
  Tim McSweeney \\
  Anglia Ruskin University \\
   Bishop Hall Ln, Chelmsford\\
  CM1 1SQ, United Kingdom \\
  \texttt{Timothy.McSweeney@aru.ac.uk}\\}

%\author{
%  \textbf{First Author\textsuperscript{1}},
%  \textbf{Second Author\textsuperscript{1,2}},
%  \textbf{Third T. Author\textsuperscript{1}},
%  \textbf{Fourth Author\textsuperscript{1}},
%\\
%  \textbf{Fifth Author\textsuperscript{1,2}},
%  \textbf{Sixth Author\textsuperscript{1}},
%  \textbf{Seventh Author\textsuperscript{1}},
%  \textbf{Eighth Author \textsuperscript{1,2,3,4}},
%\\
%  \textbf{Ninth Author\textsuperscript{1}},
%  \textbf{Tenth Author\textsuperscript{1}},
%  \textbf{Eleventh E. Author\textsuperscript{1,2,3,4,5}},
%  \textbf{Twelfth Author\textsuperscript{1}},
%\\
%  \textbf{Thirteenth Author\textsuperscript{3}},
%  \textbf{Fourteenth F. Author\textsuperscript{2,4}},
%  \textbf{Fifteenth Author\textsuperscript{1}},
%  \textbf{Sixteenth Author\textsuperscript{1}},
%\\
%  \textbf{Seventeenth S. Author\textsuperscript{4,5}},
%  \textbf{Eighteenth Author\textsuperscript{3,4}},
%  \textbf{Nineteenth N. Author\textsuperscript{2,5}},
%  \textbf{Twentieth Author\textsuperscript{1}}
%\\
%\\
%  \textsuperscript{1}Affiliation 1,
%  \textsuperscript{2}Affiliation 2,
%  \textsuperscript{3}Affiliation 3,
%  \textsuperscript{4}Affiliation 4,
%  \textsuperscript{5}Affiliation 5
%\\
%  \small{
%    \textbf{Correspondence:} \href{mailto:email@domain}{email@domain}
%  }
%}

\begin{document}
\maketitle
\begin{abstract}
This paper presents an analysis of ten years (2013–-2023) of magistrates’ court data on child sexual abuse (CSA) prosecutions in England and Wales. A dataset comprising 69,685 cases was examined to identify trends in prosecution volumes, offence classifications, defendant counts, and the geographical distribution of cases. Fourteen Home Office offence-group codes were mapped to the published Home Office/VKPP offence classification to ensure accurate interpretation of the legal offence categories.

The findings indicate that the number of prosecuted CSA cases increased steadily until reaching a peak in 2017, followed by a marked decline that began two years before the COVID-19 pandemic, suggesting that the decrease cannot be attributed solely to pandemic-related disruption. The analysis also identifies a notable geographical pattern: unlike crime in general, prosecuted CSA cases are relatively dispersed across neighbourhoods rather than concentrated within a small number of locations. This unexpected finding suggests that the spatial distribution of prosecuted CSA differs from that of many other offence types. Taken together, the temporal and geographical patterns highlight the value of magistrates’ court records as a source of evidence for understanding prosecuted CSA and contribute to ongoing discussions about the scale, distribution, and measurement of child sexual abuse in England and Wales.


\end{abstract}

\section{Introduction}

Current evidence on child sexual abuse (CSA) in England and Wales is derived primarily from two sources. The first is police-recorded crime statistics, published most comprehensively through the Vulnerability Knowledge and Practice Programme (VKPP) in its annual national analyses \citep{vkpp2025codes}. The second comprises survey-based prevalence studies that seek to estimate the extent of abuse that remains undisclosed and therefore absent from official records. Comparatively little attention has been given to the intermediate stage of the criminal justice process: cases that have been recorded by the police and progressed to the magistrates’ courts, where almost all criminal proceedings in England and Wales begin, including those later committed to the Crown Court.

This paper addresses that gap by analysing linked magistrates’ court data covering the period from 2013 to 2023. The analysis examines annual prosecution volumes, offence types, the number of defendants charged per case, and the geographical distribution of cases across multiple spatial scales, from regional patterns to neighbourhood-level variation. To ensure consistent interpretation of offences, the Home Office offence-group codes contained within the dataset were mapped to the published Home Office/VKPP offence classification for CSA-related offences \citep{vkpp2025codes}.

A substantial proportion of the offences examined are directly associated with online activity. Indecent imagery offences account for more than one-third of all prosecuted cases, while grooming and sexual communication offences, although less prevalent, are inherently facilitated through online interaction. The dataset therefore provides a perspective on the criminal justice outcomes of online harms that are more commonly examined by the Workshop on Online Abuse and Harms (WOAH) research community at the point of detection.

The study period closely aligns with the first decade of the WOAH workshop series, making it possible to compare changes in prosecuted CSA offences with the broader evolution of online harms documented over the same period. Examining these long-term trends provides additional insight into how changes in offending patterns, legislation, and criminal justice practice have shaped CSA prosecutions during the past decade.



\section{Related Work}
Three strands of literature provide the context for this study, although they have received relatively little attention in combination. First, the annual reports produced by the Vulnerability Knowledge and Practice Programme (VKPP) describe police-recorded CSA offences prior to charging decisions \citep{vkpp2023report}, \citep{vkpp2024report}, whereas the present analysis focuses on cases that have progressed into the court system. Second, criminological and psychological research has examined the characteristics of grooming, offending behaviour, and the demographic profiles of children referred to exploitation services, including variation by gender, age, and geography \citep{whittle2013reviewgrooming,cockbain2017immaterial}. Third, research on the geography of crime has demonstrated that recorded crime patterns are shaped not only by the distribution of offending but also by reporting behaviour and policing practices \citep{ashby2019crime}, an important consideration when interpreting the spatial distribution of prosecuted CSA cases.

Alongside these strands, a substantial body of computational research has focused on the detection of online grooming and child sexual abuse within digital communications \citep{borj2022online}. However, comparatively little attention has been given to the progression of these offences beyond the point of detection and into the criminal justice system \citep{vidgen2019challenges}. By examining prosecuted CSA cases over a ten-year period, the present study contributes empirical evidence on this later stage of the offence pathway.


\section{Data and Methods}
The analysis was based on magistrates’ court records from England and Wales covering the period 2013–-2023. The dataset includes annual counts of child sexual abuse (CSA) cases by Home Office offence code, cumulative offence totals across the study period, defendant counts, and geographical information at regional, local authority, and Lower Layer Super Output Area (LSOA) levels. Corresponding data for all recorded offences were also included to provide a population-level baseline for comparison. Home Office offence codes were mapped to the published VKPP/Home Office offence classification to ensure consistent interpretation of offence categories \citep{vkpp2025codes}. Local authority identifiers follow the Office for National Statistics (ONS) coding scheme. In accordance with UK statistical disclosure controls, counts below 10 are suppressed and reported as “<10”.

\section{Results}

%\subsection{Number of cases}
Across the study period, the dataset records 69,685 CSA cases spanning fourteen Home Office offence-group codes (Table~\ref{tab:group}). Indecent imagery of children constitutes the largest offence category, accounting for 27,135 cases, or more than one-third of the total. Sexual assault of a young child and sexual activity with a child under 16 together account for a further 20,484 cases. Grooming-related offences represent a substantially smaller proportion of prosecuted cases (3,586), although they exhibit one of the strongest upward trends over the study period.

\begin{table}
  \centering
 \small
  %\begin{tabular}{lc}
 %   \hline
%\begin{table}[t]
%\centering
%\small
\begin{tabular}{lrl}
\hline
%\toprule
\textbf{Cases} &\textbf{Description}\\
    \hline
%Group & Cases & What it means \\
%\midrule
27{,}135 & Indecent images (non-contact) \\
 10{,}604 & Sexual assault, child u13 \\
 9{,}880 & Sexual activity, child u16 \\
 6{,}853 & Rape of a child u16 \\
 3{,}586 & Sexual grooming/communication \\
 3{,}198 & Rape of a child u13 \\
 3{,}125 & Sexual activity, child u13 \\
1{,}635 & Sexual exploitation of a child \\
 933 & Sexual assault, male child u13 \\
 891 & Rape, male child u13 \\
 664 & Abuse, family member \\
 599 & Rape, male child u16 \\
393 & Rape, male aged 16+ \\
 189 & Abuse of trust \\
%\midrule
69{,}685 & Total&\\
\hline
%\bottomrule
\end{tabular}
\caption{Home Office child sexual abuse offence-group codes included in the dataset, 2013–-2023, mapped to VKPP/Home Office offence descriptions \citep{vkpp2025codes}.}
\label{tab:group}
\end{table}


Annual prosecution volumes increased from 6,171 cases in 2013 to a peak of 8,295 in 2017 before declining sharply to 5,840 cases in 2018 (Figure~\ref{fig:experiments}), a reduction of almost 30\% that occurred two years before the COVID-19 pandemic. Prosecution volumes did not subsequently return to their 2017 level. At the individual offence-code level, Home Office code 8602 (“take, make or distribute indecent images of children”) was consistently the largest contributor throughout the study period, accounting for 24,059 cases. In contrast, grooming-specific offences, including sexual communication with a child and meeting a child following grooming, were rarely recorded before 2017 but increased steadily thereafter, suggesting changes in charging practices and legal classification over time rather than necessarily reflecting changes in offending prevalence.
  



\begin{figure*}[t]
 % \includegraphics[width=\columnwidth]{figtrend.pdf}
 \centering
 \includegraphics[width=0.70\linewidth]{Figure1.pdf} 
  \caption{Trends in annual CSA prosecution volume (left axis) and single-defendant sexual offence cases (right axis) from 2013 to 2023. The shaded region marks the onset of the COVID-19 pandemic. Figures for 2023 represent a partial year.}
  \label{fig:experiments}
\end{figure*}

%\subsection{Number of defendant }
Most prosecuted cases involved a single defendant, reflecting the person-level structure of the magistrates’ court dataset. The pattern observed closely mirrors broader trends in sexual offence prosecutions, with increases until the middle of the decade followed by a marked decline from 2018, partial recovery in 2020, and a further decrease by 2022 (Figure~\ref{fig:experiments}).


%\subsection{Geographically distribution of the defendants}
Regional variation in prosecuted CSA cases does not correspond directly to population size or overall crime volume. Although London recorded the largest number of offences overall (2,347,843), it ranked fourth for CSA prosecutions (6,810), behind the North West (10,132), the South East (8,132), and Yorkshire and The Humber (7,352). Consequently, CSA represented only 0.29\% of all recorded offences in London, compared with 0.62\% in the South West. This difference appears to reflect London’s substantially larger volume of recorded crime overall rather than indicating a lower prevalence of CSA.
At the local authority level, the highest numbers of prosecuted CSA cases were recorded in large urban authorities. Birmingham recorded the largest total (1,178 cases), followed by Leeds, Manchester, County Durham, and Bradford (Table~\ref{tab:la}). In contrast, the lowest disclosed counts were observed in smaller and more rural authorities, including Perth and Kinross (11 cases), Dumfries and Galloway (12), Rutland (29), and several other sparsely populated authorities. A further 38 local authorities recorded fewer than ten cases across the study period and are not reported individually because of disclosure controls.



\begin{table}[t]
\centering
\small
\begin{tabular}{lr}
\hline
%\multicolumn{2}{l}{\textit{Busiest 10 (CSA cases)}} \\
\textbf{Busiest 10 (CSA cases)} & \textbf{Number of Cases} \\
    \hline
Birmingham & 1{,}178 \\
Leeds & 1{,}034 \\
Manchester & 805 \\
County Durham & 794 \\
Bradford & 791 \\
Liverpool & 681 \\
Kirklees & 666 \\
Wakefield & 566 \\
Sheffield & 544 \\
Doncaster & 485 \\
\hline
%\multicolumn{2}{l}{\textit{Quietest 10 (CSA cases, excl.\ suppressed)}} \\
\textbf{Quietest 10 (CSA cases)} & \textbf{Number of Cases} \\
    \hline
    Perth and Kinross      & 11 \\
Dumfries \& Galloway & 12 \\
\multicolumn{2}{l}{8 further small authorities} \\
\multicolumn{2}{l}{\quad (codes only; 13--37 cases each)} \\
%\bottomrule
\hline
\end{tabular}
\caption{Local authorities with the highest and lowest numbers of prosecuted CSA cases, 2013–2023. Authorities with fewer than 10 cases are omitted in accordance with disclosure controls.}
\label{tab:la}
\end{table}

The most notable finding emerges at neighbourhood level. General crime exhibits substantial spatial concentration, with the busiest 10\% of Lower Layer Super Output Areas (LSOAs) accounting for 31\% of all recorded offences. Prosecuted CSA cases display a markedly different distribution: the busiest 10\% of neighbourhoods account for only 12\% of CSA cases, while the busiest 1\% account for just 3.2\% (Figure~\ref{fig:lorenz}). Unlike recorded crime more broadly, prosecuted CSA appears to be comparatively dispersed rather than concentrated within a small number of geographic hotspots. This finding challenges the assumption that the spatial distribution of CSA resembles that of other offence types and has important implications for policing, safeguarding, and the geographical targeting of prevention and intervention strategies.


\begin{figure}[t]
\centering
\includegraphics[width=0.80\columnwidth]{Figure2.pdf}
\caption{Cumulative distribution of prosecuted CSA cases and all recorded offences across neighbourhoods (LSOAs), ranked by case volume. The dashed line indicates perfect spatial equality. General crime is more spatially concentrated than prosecuted CSA.}
\label{fig:lorenz}
\end{figure}

\section{Discussion}
The decline in prosecuted CSA cases after 2017 should be interpreted with caution. Although the reduction may appear to indicate a decrease in offending, prosecution volumes are influenced by multiple stages of the criminal justice process, including victim disclosure, police recording practices, charging decisions, prosecutorial thresholds, and court capacity. Several of these factors changed during the study period for reasons unrelated to the underlying prevalence of abuse. This interpretation is further supported by VKPP statistics, which show that police-recorded CSA offences continued to increase over a similar period \citep{vkpp2024report}. The divergence between rising police-recorded offences and declining prosecutions highlights an important area for future investigation that cannot be fully addressed using court data alone.

The offence-level analysis also has implications for computational research. The concentration of prosecutions within a relatively small number of offence categories provides a practical legal taxonomy that can inform the development and evaluation of CSA detection systems. However, the steady increase in grooming-specific offence codes after 2017 demonstrates that legal classifications evolve over time. Consequently, detection models trained using historical offence categories may encode assumptions that no longer reflect current charging practice. Rather than treating offence taxonomies as fixed ground truth, they should be reviewed and updated periodically to ensure that datasets and detection systems remain aligned with contemporary legal definitions and prosecutorial practice.

The geographical findings present a different challenge to existing assumptions. Unlike recorded crime more broadly, prosecuted CSA cases exhibit relatively limited spatial concentration at neighbourhood level. One possible explanation is that disclosure is driven primarily by individual and institutional factors, such as family circumstances or intervention by schools and safeguarding professionals, rather than by neighbourhood characteristics commonly associated with other forms of crime. Alternatively, many forms of CSA, particularly those facilitated through online communication, may be less dependent on physical location than offences such as burglary or street violence. If so, approaches to safeguarding that rely primarily on geographically targeted interventions may be less effective than strategies centred on victims, offenders, and online environments.





\section{Conclusion}
Analysis of ten years of magistrates’ court data identifies several important patterns in prosecuted child sexual abuse (CSA) cases in England and Wales. Prosecution volumes increased until 2017 before declining prior to the COVID-19 pandemic and prosecutions were concentrated within a relatively small number of legal offence categories despite notable changes in charging practices over time. Spatial analysis further demonstrated that, unlike recorded crime more generally, prosecuted CSA cases are comparatively dispersed rather than concentrated within a small number of geographic hotspots.

These findings should not be interpreted as evidence of changes in the underlying prevalence of CSA, as prosecution data reflect only one stage of the criminal justice process and are influenced by reporting, investigation, charging, and court procedures. Nevertheless, magistrates’ court records provide a valuable and comparatively underused source of evidence for understanding how CSA progresses through the justice system.

By linking prosecution data to established Home Office offence classifications and examining long-term temporal and geographical patterns, this study contributes empirical evidence that is relevant to researchers, policymakers, safeguarding practitioners, and the online harms community. The findings also highlight the importance of interpreting court data within the broader context of changing legal classifications and criminal justice practices when analysing trends in child sexual abuse.



\section*{Limitations}
Several limitations should be considered when interpreting these findings. In accordance with UK statistical disclosure controls, counts below 10 were suppressed in the source data. For visualisation purposes only, suppressed values were replaced with a value of five in figures; these substitutions were not used in summary statistics or reported totals. As a result, analyses involving smaller local authorities and the devolved nations of Wales, Scotland, and Northern Ireland are more affected by disclosure suppression than those for larger areas.

The 2023 data represent a partial year and should therefore be interpreted with caution. In addition, although most Home Office offence codes were successfully mapped to published legal classifications, a small number could not be confidently identified and were consequently left unclassified. Finally, magistrates’ court records capture only cases that progressed to prosecution and do not include offences resolved outside the criminal courts. Consequently, prosecution volumes are substantially lower than the police-recorded CSA figures reported by the Vulnerability Knowledge and Practice Programme (VKPP) for the same period \citep{vkpp2024report}. The results presented are descriptive and should not be interpreted as evidence of causal relationships.


\section*{Ethics Statement}
All data analysed in this study were provided in pre-aggregated form, with small counts suppressed prior to access. Consequently, no individual victim, defendant, or case can be identified from the data or the results presented. Given the sensitive nature of child sexual abuse research, the analysis was limited to descriptive aggregate statistics, and no inferences were made regarding the role of specific platforms, technologies, or other factors that were not directly supported by the available evidence. The study is intended to contribute to evidence-based research on child protection, online harms, and the operation of the criminal justice system.


\bibliographystyle{plainnat}
\bibliography{refs}

\end{document}

%@techreport{vkpp2023report,
  title={National Analysis of Police-Recorded Child Sexual Abuse and Exploitation Crimes Report 2023},
  author={{Vulnerability Knowledge and Practice Programme (VKPP)}},
  institution={National Police Chiefs' Council / College of Policing},
  year={2023},
  url={https://www.vkpp.org.uk/vkpp-work/analytical-capability/national-analysis-of-police-recorded-child-sexual-abuse-and-exploitation-crimes-report-2023/}
}

\section{Engines}

To produce a PDF file, pdf\LaTeX{} is strongly recommended (over original \LaTeX{} plus dvips+ps2pdf or dvipdf).
The style file \texttt{acl.sty} can also be used with
lua\LaTeX{} and
Xe\LaTeX{}, which are especially suitable for text in non-Latin scripts.
The file \texttt{acl\_lualatex.tex} in this repository provides
an example of how to use \texttt{acl.sty} with either
lua\LaTeX{} or
Xe\LaTeX{}.

\section{Preamble}

The first line of the file must be
\begin{quote}
\begin{verbatim}
\documentclass[11pt]{article}
\end{verbatim}
\end{quote}

To load the style file in the review version:
\begin{quote}
\begin{verbatim}
\usepackage[review]{acl}
\end{verbatim}
\end{quote}
For the final version, omit the \verb|review| option:
\begin{quote}
\begin{verbatim}
\usepackage{acl}
\end{verbatim}
\end{quote}

To use Times Roman, put the following in the preamble:
\begin{quote}
\begin{verbatim}
\usepackage{times}
\end{verbatim}
\end{quote}
(Alternatives like txfonts or newtx are also acceptable.)

Please see the \LaTeX{} source of this document for comments on other packages that may be useful.

Set the title and author using \verb|\title| and \verb|\author|. Within the author list, format multiple authors using \verb|\and| and \verb|\And| and \verb|\AND|; please see the \LaTeX{} source for examples.

By default, the box containing the title and author names is set to the minimum of 5 cm. If you need more space, include the following in the preamble:
\begin{quote}
\begin{verbatim}
\setlength\titlebox{<dim>}
\end{verbatim}
\end{quote}
where \verb|<dim>| is replaced with a length. Do not set this length smaller than 5 cm.

\section{Document Body}

\subsection{Footnotes}

Footnotes are inserted with the \verb|\footnote| command.\footnote{This is a footnote.}

\subsection{Tables and figures}

See Table~\ref{tab:accents} for an example of a table and its caption.
\textbf{Do not override the default caption sizes.}

\begin{table}
  \centering
  \begin{tabular}{lc}
    \hline
    \textbf{Command} & \textbf{Output} \\
    \hline
    \verb|{\"a}|     & {\"a}           \\
    \verb|{\^e}|     & {\^e}           \\
    \verb|{\`i}|     & {\`i}           \\
    \verb|{\.I}|     & {\.I}           \\
    \verb|{\o}|      & {\o}            \\
    \verb|{\'u}|     & {\'u}           \\
    \verb|{\aa}|     & {\aa}           \\\hline
  \end{tabular}
  \begin{tabular}{lc}
    \hline
    \textbf{Command} & \textbf{Output} \\
    \hline
    \verb|{\c c}|    & {\c c}          \\
    \verb|{\u g}|    & {\u g}          \\
    \verb|{\l}|      & {\l}            \\
    \verb|{\~n}|     & {\~n}           \\
    \verb|{\H o}|    & {\H o}          \\
    \verb|{\v r}|    & {\v r}          \\
    \verb|{\ss}|     & {\ss}           \\
    \hline
  \end{tabular}
  \caption{Example commands for accented characters, to be used in, \emph{e.g.}, Bib\TeX{} entries.}
  \label{tab:accents}
\end{table}

As much as possible, fonts in figures should conform
to the document fonts. See Figure~\ref{fig:experiments} for an example of a figure and its caption.

Using the \verb|graphicx| package graphics files can be included within figure
environment at an appropriate point within the text.
The \verb|graphicx| package supports various optional arguments to control the
appearance of the figure.
You must include it explicitly in the \LaTeX{} preamble (after the
\verb|\documentclass| declaration and before \verb|\begin{document}|) using
\verb|\usepackage{graphicx}|.

\begin{figure}[t]
  \includegraphics[width=\columnwidth]{example-image-golden}
  \caption{A figure with a caption that runs for more than one line.
    Example image is usually available through the \texttt{mwe} package
    without even mentioning it in the preamble.}
  \label{fig:experiments}
\end{figure}

\begin{figure*}[t]
  \includegraphics[width=0.48\linewidth]{example-image-a} \hfill
  \includegraphics[width=0.48\linewidth]{example-image-b}
  \caption {A minimal working example to demonstrate how to place
    two images side-by-side.}
\end{figure*}

\subsection{Hyperlinks}

Users of older versions of \LaTeX{} may encounter the following error during compilation:
\begin{quote}
\verb|\pdfendlink| ended up in different nesting level than \verb|\pdfstartlink|.
\end{quote}
This happens when pdf\LaTeX{} is used and a citation splits across a page boundary. The best way to fix this is to upgrade \LaTeX{} to 2018-12-01 or later.

\subsection{Citations}

\begin{table*}
  \centering
  \begin{tabular}{lll}
    \hline
    \textbf{Output}           & \textbf{natbib command} & \textbf{ACL only command} \\
    \hline
    \citep{Gusfield:97}       & \verb|\citep|           &                           \\
    \citealp{Gusfield:97}     & \verb|\citealp|         &                           \\
    \citet{Gusfield:97}       & \verb|\citet|           &                           \\
    \citeyearpar{Gusfield:97} & \verb|\citeyearpar|     &                           \\
    \citeposs{Gusfield:97}    &                         & \verb|\citeposs|          \\
    \hline
  \end{tabular}
  \caption{\label{citation-guide}
    Citation commands supported by the style file.
    The style is based on the natbib package and supports all natbib citation commands.
    It also supports commands defined in previous ACL style files for compatibility.
  }
\end{table*}

Table~\ref{citation-guide} shows the syntax supported by the style files.
We encourage you to use the natbib styles.
You can use the command \verb|\citet| (cite in text) to get ``author (year)'' citations, like this citation to a paper by \citet{Gusfield:97}.
You can use the command \verb|\citep| (cite in parentheses) to get ``(author, year)'' citations \citep{Gusfield:97}.
You can use the command \verb|\citealp| (alternative cite without parentheses) to get ``author, year'' citations, which is useful for using citations within parentheses (e.g. \citealp{Gusfield:97}).

A possessive citation can be made with the command \verb|\citeposs|.
This is not a standard natbib command, so it is generally not compatible
with other style files.

\subsection{References}

\nocite{Ando2005,andrew2007scalable,rasooli-tetrault-2015}

The \LaTeX{} and Bib\TeX{} style files provided roughly follow the American Psychological Association format.
If your own bib file is named \texttt{custom.bib}, then placing the following before any appendices in your \LaTeX{} file will generate the references section for you:
\begin{quote}
\begin{verbatim}
\bibliography{custom}
\end{verbatim}
\end{quote}

You can obtain the complete ACL Anthology as a Bib\TeX{} file from \url{https://aclweb.org/anthology/anthology.bib.gz}.
To include both the Anthology and your own .bib file, use the following instead of the above.
\begin{quote}
\begin{verbatim}
\bibliography{anthology,custom}
\end{verbatim}
\end{quote}

Please see Section~\ref{sec:bibtex} for information on preparing Bib\TeX{} files.

\subsection{Equations}

An example equation is shown below:
\begin{equation}
  \label{eq:example}
  A = \pi r^2
\end{equation}

Labels for equation numbers, sections, subsections, figures and tables
are all defined with the \verb|\label{label}| command and cross references
to them are made with the \verb|\ref{label}| command.

This an example cross-reference to Equation~\ref{eq:example}.

\subsection{Appendices}

Use \verb|\appendix| before any appendix section to switch the section numbering over to letters. See Appendix~\ref{sec:appendix} for an example.

\section{Bib\TeX{} Files}
\label{sec:bibtex}

Unicode cannot be used in Bib\TeX{} entries, and some ways of typing special characters can disrupt Bib\TeX's alphabetization. The recommended way of typing special characters is shown in Table~\ref{tab:accents}.

Please ensure that Bib\TeX{} records contain DOIs or URLs when possible, and for all the ACL materials that you reference.
Use the \verb|doi| field for DOIs and the \verb|url| field for URLs.
If a Bib\TeX{} entry has a URL or DOI field, the paper title in the references section will appear as a hyperlink to the paper, using the hyperref \LaTeX{} package.

\section*{Limitations}

This document does not cover the content requirements for ACL or any
other specific venue.  Check the author instructions for
information on
maximum page lengths, the required ``Limitations'' section,
and so on.

\section*{Acknowledgments}

This document has been adapted
by Steven Bethard, Ryan Cotterell and Rui Yan
from the instructions for earlier ACL and NAACL proceedings, including those for
ACL 2019 by Douwe Kiela and Ivan Vuli\'{c},
NAACL 2019 by Stephanie Lukin and Alla Roskovskaya,
ACL 2018 by Shay Cohen, Kevin Gimpel, and Wei Lu,
NAACL 2018 by Margaret Mitchell and Stephanie Lukin,
Bib\TeX{} suggestions for (NA)ACL 2017/2018 from Jason Eisner,
ACL 2017 by Dan Gildea and Min-Yen Kan,
NAACL 2017 by Margaret Mitchell,
ACL 2012 by Maggie Li and Michael White,
ACL 2010 by Jing-Shin Chang and Philipp Koehn,
ACL 2008 by Johanna D. Moore, Simone Teufel, James Allan, and Sadaoki Furui,
ACL 2005 by Hwee Tou Ng and Kemal Oflazer,
ACL 2002 by Eugene Charniak and Dekang Lin,
and earlier ACL and EACL formats written by several people, including
John Chen, Henry S. Thompson and Donald Walker.
Additional elements were taken from the formatting instructions of the \emph{International Joint Conference on Artificial Intelligence} and the \emph{Conference on Computer Vision and Pattern Recognition}.

% Bibliography entries for the entire Anthology, followed by custom entries
%\bibliography{custom,anthology-overleaf-1,anthology-overleaf-2}

% Custom bibliography entries only
\bibliography{custom}

\appendix

\section{Example Appendix}
\label{sec:appendix}

This is an appendix.

\end{document}