[AISWorld] Challenge at WWW 2018: Financial Opinion Mining and Question Answering

[André Freitas]

Financial Opinion Mining and Question Answering

Open Challenge - WWW 2018 LYON, FRANCE (23 - 27 April 2018)

Website: https://sites.google.com/view/fiqa

Summary

The growing maturity of Natural Language Processing (NLP) techniques and
resources is drastically changing the landscape of many application domains
which are dependent on the analysis of unstructured data at scale. The
financial domain, with its dependency on the interpretation of multiple
unstructured and structured data sources and with its demand for fast and
comprehensive decision making is already emerging as a primary ground for
the experimentation of NLP, Web Mining and Information Retrieval (IR)
techniques. This challenge focuses on advancing the state-of-the-art of
aspect-based sentiment analysis and opinion-based Question Answering for
the financial domain.

Topics of particular interest to be discussed and developed within the task
include (but are not limited to):

* Aspect-oriented sentiment analysis and opinion mining.
* Aspect-identification extraction/classification for finance for opinion
mining.
* Question Answering and opinion-based Question Answering over financial
text.
* Multi-lingual sentiment analysis.
* Linguistic analysis tools for the financial domain, in particular
financial social media (e.g. tokenisation, part-of-speech tagging,
normalization,  parsing)
* Sentiment classification on financial texts;
* Analysing and understanding  linguistic phenomena associated with
financial text corpora (including the sub-language of financial microblogs);
* New semantic and ontological models for finance;
* Construction and application of distributional semantic models on finance;
* Lexical resources for the financial domain;

Tasks

Two tasks will be available to participating systems (participants can be
involved in one of the tasks or both):

Task 1: Aspect-based financial sentiment analysis

Given a text instance in the financial domain (microblog message, news
statement or headline) in English, detect the target
companies/commodities/currencies which are mentioned in the text and
predict the sentiment score for each of the mentioned targets. Sentiment
values will be defined using a 5 point discrete scale: very bearish
(negative), bearish, neutral, bullish, very bullish.

Task 2: Opinion-based QA over financial data

Given a corpus of structured and unstructured text documents from different
financial data sources in English (microblogs, reports, news) build a
Question Answering system that answers natural language questions. For this
challenge, part of the questions will be opinionated, targeting mined
opinions and their respective entities, aspects, sentiment polarity and
opinion holder.

Challenge Timeline

* Release of the training data : December 5th, 2017
* Challenge papers submission deadline : February 4th, 2018
* Challenge papers acceptance notification : February 14th, 2018
* Challenge test data published : February 14th, 2018

Organizers

André Freitas, School of Computer Science, University of Manchester.
Alexandra Balahur, Text and Data Mining Unit, European Commission's Joint
Research Centre.
Manel Zarrouk, Insight Centre for Data Analytics, National University of
Ireland, Galway.
Macedo Maia, Department of Computer Science and Mathematics, University of
Passau.
Brian Davis, Department of Computer Science, Maynooth University.