Anthology development and API

Information on how to programmatically access Anthology data

November 15, 2019


All Anthology data is stored under the data directory. In the xml directory are the files that contain all the Anthology metadata, in a format described below. The yaml directory contains various other important information.

In addition, we have a Python API that defines objects for papers, authors, volumes, and so on. This is all accessible by an Anthology object that can be access in the following manner. First, clone the Anthology Github repo and setup your environment:

git clone
export ACLANTHOLOGY=$(pwd)/acl-anthology

You can now access the Anthology from Python:

from anthology import Anthology

anthology = Anthology(importdir='acl-anthology/data')
for id_, paper in anthology.papers.items():
  print(paper.anthology_id, paper.get_title('text')

See the script to see how to iterate over the Anthology object and access other information such as authors and volumes.

Authoritative XML format

The Anthology site is generated from an authoritative XML file format containing information about volumes, paper titles, and authors. This data is stored in the official repository on Github. Here is a fragment of a complete XML file (P18.xml), to give you the idea. The full file contains much more information.

<?xml version="1.0" encoding="UTF-8" ?>
<collection id="P18">
  <volume id="3">
      <booktitle>Proceedings of <fixed-case>ACL</fixed-case> 2018, Student Research Workshop</booktitle>
      <!-- ... -->
    <paper id="1">
      <title>Towards Opinion Summarization of Customer Reviews</title>
      <!-- ... -->
    <paper id="2">
      <title>Sampling Informative Training Data for <fixed-case>RNN</fixed-case> Language Models</title>
      <!-- ... -->
    <paper id="3">
      <title>Learning-based Composite Metrics for Improved Caption Evaluation</title>
      <author><first>Syed Afaq</first><last>Ali Shah</last></author>
      <!-- ... -->
    <!-- ...  -->

Our scripts use the library to parse the XML. You can see examples of parsing and accessing the XML directly in and