CFP: Shared Task on Information Extractor for Conversational Systems in Indian Languages - FIRE' 18

[barathi ganesh hb]

In today's digital world, supplying digital solutions using Artificial Intelligence through mobile applications or web services have become essential. Conversational Systems is one such application, that tries to assist the end-users in every domain where there exist human - human communication. The Conversational and Personal Assistant Systems tries to replace the human - human communication by machine - human communication. For example, replacing human efforts involved in customer care unit in an organisation with Conversational Systems (chat bot).

Building an intelligent Conversational System that performs closer to human is still in research. The existing systems are still focusing on the betterment of its outcome and performs better with the universal language English (till date). India with a population count of around 1324 million, results in high potential market opportunity for industries and it has more than 20 scheduled languages that reflects the high complexity involved in building an Indian language based Conversational System. Building a Natural Language Understander (NLU), which is the core part of any Conversational System for Indian Language is essential and also highly complex to develop.

By understanding these prerequisites, ARNEKT Solutions Pvt. Ltd. in conjunction with Forum for Information Retrieval and Evaluation (FIRE 2018) is hosting a shared task to motivate the researchers and students to develop, one of the core NLU component, known as Information Extractor for five Indian Languages (Hindi, Tamil, Malayalam, Telugu and Kannada). We believe this initiative will move towards building a successful Indian language based Conversational System.

We are glad to invite researchers/ research groups and students to participate in the Shared Task on Information Extractor for Conversational Systems in Indian Languages (ARNEKT-IECSIL) in conjunction with Forum for Information Retrieval Evaluation (FIRE) 2018. Detailed description about the task and other key points are given here.

Information Extractor for Conversational Systems in Indian Languages (ARNEKT-IECSIL)

Motivated by the need for Information Extractor, ARNEKT-IECSIL have the following two tasks:

Task A : Named Entity Recognition (NER)

Corpora for five Indian languages (Hindi, Tamil, Malayalam, Telugu and Kannada) would be provided. Task A is to identify and classify the named entities to one of the many classes.

Task B : Relation Extraction (RE)

Continuation to Task A, corpora with named entities for five Indian languages (Hindi, Tamil, Malayalam, Telugu and Kannada) would be provided. Task B is to extract the relation amongst the entities provided.

Schedule

May 1    : Registration open

May 15  : Train corpora release

May 20  : Test-1 corpora release

July 15  : Test-2 corpora release

July 15  : Registration close

July 20  : Run submission deadline

July 25  : Results declared

July 30  : Working notes submission

Aug 10  : Camera ready submission of working notes

Participate in ARNEKT-IECSIL shared task to claim your chance of glory – the best people will be shortlisted for an interview process with ARNEKT Solutions Pvt. Ltd, Pune. To top it up, the top three teams would take exciting prizes with them and all the participants will be awarded with certificates.