Subfield survey
Understanding how to read and synthesize articles in NLP is an important part of carrying out research in this space. To cultivate this skill, your final report will be a survey (2000 words [+/- 200 words], excluding the list of references at the end) for a specific NLP subfield of your choice (e.g., coreference resolution, question answering, interpretability, narrative generation, etc.), synthesizing at least 25 papers published in NLP or related venues; 10 of these citations must be from ACL, EMNLP, NAACL, EACL, AACL, CoNLL, Transactions of the ACL or Computational Linguistics (the main conferences and journals, not affiliated workshops); the remaining 15 can be from other venues so long as they are computational in nature (e.g., ICML, NeurIPS, ICLR, ArXiv, any workshops associated with the main conferences above). Your survey should report the word count as a footnote on the first page.
This survey should be able to provide a newcomer (such as yourself at the start of the semester) the following information:
A sense of the current state of the art in the subfield in 2023
The major historical papers that have defined the area
Any different schools of thought or classes of solutions within it. (An example of different schools of thought would be the different approaches to machine translation discussed in SLP3 ch. 13 (pp. 19-21), or the difference between generative, discriminative, and neural approaches we’ve seen throughout the semester).
Your survey should not just be a list of 25 bullet points describing each paper in turn; you should take care to synthesize all of the works into a continuous narrative well enough that your survey conveys a comprehensive view of the subfield, identifying the major tasks, datasets, methods, and means of evaluation. See the example surveys below for the kind of description (but not the length!) we expect. Importantly, every paper you cite must include a URL to the source.
The survey should use the ACL 2023 style files for formatting, which are available as an Overleaf template. (Links to an external site.).
To calculate word count on Overleaf, see this post: https://www.overleaf.com/blog/word-count-2015-09-15 (Links to an external site.). For an overview on how to use overleaf, see this tutorial (Links to an external site.).
Here are some existing surveys — while these are all of course much longer than the ones you will be writing, they can still give you a sense about what the genre of a survey is all about:
“Automated Essay Scoring: A Survey of the State of the Art” (Links to an external site.)
“A Survey on Recent Advances in Named Entity Recognition from Deep Learning models” (Links to an external site.)
“A Survey of Arabic Named Entity Recognition and Classification” (Links to an external site.)
“A Comprehensive Survey of Deep Learning for Image Captioning” (Links to an external site.)
“A Systematic Review of Automatic Question Generation for Educational Purposes” (Links to an external site.)
“Survey of the State of the Art in Natural Language Generation” (Links to an external site.)
“Extraction and Analysis of Fictional Character Networks: A Survey” (Links to an external site.)
Submit your subfield survey as a pdf file through bCourses by 11:59pm on May 9.
Venues
Valid NLP venues. At least 10 papers must come from the main conference of only these venues (not workshops associated with them). We provide links for the most recent edition of the conference for reference.
ACL: https://aclanthology.org/volumes/2022.acl-long/Links to an external site.
EMNLP: https://aclanthology.org/events/emnlp-2022/#2022emnlp-mainLinks to an external site.
NAACL: https://aclanthology.org/events/naacl-2022/#2022naacl-mainLinks to an external site.
EACL: https://aclanthology.org/events/eacl-2021/#2021eacl-mainLinks to an external site.
AACL: https://aclanthology.org/events/aacl-2022/#2022aacl-mainLinks to an external site.
Transactions of the ACL: https://aclanthology.org/events/tacl-2022/Links to an external site.
Computational Linguistics: https://aclanthology.org/events/cl-2022/Links to an external site.
CoNLL: https://aclanthology.org/events/conll-2022/Links to an external site.
Findings of ACL, EMNLP, etc.: https://aclanthology.org/events/findings-2022/Links to an external site.
Sample computational conferences/workshops/journals.
ICML
ICLR
NeurIPS
Workshop on Statistical Machine Translation: https://aclanthology.org/events/wmt-2022/Links to an external site.
Workshop on Figurative Language Processing (co-located with EMNLP 2022): https://aclanthology.org/events/emnlp-2022/#2022flp-1Links to an external site.
Workshop on Biomedical Language Processing (co-located with ACL 2022): https://aclanthology.org/events/acl-2022/#2022bionlp-1Links to an external site.
ArXiv: https://arxiv.org/Links to an external site.
Invalid sources
Just to clarify, you should only use papers from peer-reviewed conferences, workshops and journals, and papers published on ArXiv. You should not use any other materials from the web (including blog posts, tweets, or web pages) or any material from textbooks.