TCS Academic Alliances - AsiaIndian Institute of Technology - Mumbai Indian Institute of Technology - Kanpur Indian Institute of Information Technology - Hyderabad
TCS–IITB Laboratory for Intelligent Systems TCS has established a Laboratory for Intelligent Systems (LIS) at department of Computer Science, Indian Institute of Technology, Bombay. Research at LIS covers a wide spectrum of activities ranging from adaptive platforms for fast computation to intelligent information processing tasks of which web is an indispensable component. Research in LIS is being carried out in the areas of Networks, Natural Language Processing, Program Derivation, Network Security, Machine Learning, Real Time Systems, and Databases. Some of the research work that is being carried out are listed below. Multilingual Named Entity Recognition Named Entity Recognition is an important subtask of Information Extraction, Question-Answering, Cross Lingual IR, Machine Translation and such other NLP applications. Although a large amount of work on NER has been done for English, Indian Languages have not received much attention in this respect. Under this research, IIT has focused on Hindi NER systems and has proposed an approach combining the global characteristics (like distributional parameters of words like Term In formativeness, Term Co-occurrence Statistics etc. and language cues) and local contexts (like context words, Part-Of-Speech tags of context words, gazetteers, suffix information, orthographic features etc). IndoWordnet and Sense Dismabiguation Wordnet is a large scale activity at IIT Bombay. After the development of the Hindi wordnet, other languages of India followed suit, gradually leading to the building of the IndoWordnet structure which is a linked structure of senses in multiple languages. This important lexical resource is leading to sense disambiguation systems in multiple languages. A number of Hindi WordNet related activities has been performed in this sponsorship. Principal Investigator: Prof. Pushpak Bhattacharyya Business Intelligence The primary focus has been in time series forecasting in the domain of monthly retail sales. Inputs used are sales figures from the Time Series Library and some government sources, all available on the internet. Their approach is to decompose the original series into its "natural" components - trend, seasonality and an irregular component. The group has experimented with different decomposition approaches and has used multiple forecasters for each component series – many of these are based on exponential smoothing or are various ARIMA and SARIMA models. Principal Investigator: Prof. Bernard Menzes Principal Investigators of other projects: Prof. Varsha Apte, Prof. Om Damani, Prof. G.Nagaraja, Prof. Krithi Ramamritham, Prof. S. Sudarshan
Foundation for Research in Algorithms
TCS has funded the “Foundation for Research in Algorithms” at IIT Kanpur. The aim is to catapult the nature and quality of research in the field of design and analysis of algorithms in India. This research is focused on research to aid new methods in:
It is the first of its kind of multi university center in computer science funded by private sector. The Foundation is anchored out of IIT Kanpur, and includes participation from leading researchers from other academic institutes including IITD, IITB, IISc, TIFR, CMI and IMSc. Principal Investigator: Prof. Manindra Agarwal
Indian Institute of Information Technology - HyderabadTCS has sponsored a multi-year alliance with the IIIT Hyderabad for Collaborative Technology Innovation. IIIT Hyderabad has been engaged in pioneering work in Natural languages and Video processing. The Language Technologies Research Centre plays a key-role in development of new tools and techniques in NLP. The sponsored research is focused in the following areas. Task oriented dialog management A dialogue system is a machine capable of interacting with a human to accomplish a task in a domain. The architecture IIIT is working on is a combination of Plans and Frames architecture. Plans are used to model higher goals in a domain and templates for the preliminary goals in a domain. Traditional Frame-based architectures consider user utterance in isolation which doesn’t give any information about intentions of the user. So combining the two architectures would help in relating the utterance with the history of dialogues and explains the user intentions. Accessing Unstructured, Semi-structured Information in Enterprise Prior to the collaboration with TCS on Enterprise search area, IIIT has conducted research on role personalization for Enterprise Search and had a basic implementation of the system. In an Enterprise, role describes a set of guidelines pertaining to the work profile; this role implicitly defines the official search needs in the hierarchy. In this research sponsorship IIIT has embedded the role based approach into the two basic aspects of information retrieval system, the indexing process and the search cycle (query, index search and presentation of results). The user query is expanded using role based thesaurus, obtained from co-occurrence statistics of tagged documents. A naive Bayesian classifier is used to generate role relevant scores for re-ranking of these documents. Further, they are trying to extend existing system by modifying the search process on many fronts like: Building ontology network using existing resources, Using deeper NLP Techniques in search, Automatic Hyper linking, and Harnessing encyclopedic knowledge for enterprise search. Principal Investigators: Prof. Lakshmi Bai, Prof. Rajeev Sangal, Prof. Vasudeva Varma |
Google
Facebook
Twitter
del.icio.us
Blogger
Rain Concert
TCS is a member of the Tata Group 