Audio enabled information extraction system for cricket and hockey domains

TL;DR

This paper presents an audio-based information retrieval system tailored for cricket and hockey, utilizing voice commands to search, categorize, and summarize relevant web documents with probabilistic methods, delivering results in audio form.

Contribution

It introduces a novel voice-activated system that combines concept and keyword search with probabilistic summarization for sports domain information retrieval.

Findings

Effective voice command processing for sports queries

Accurate summarization of retrieved information

Audio output enhances user accessibility

Abstract

The proposed system aims at the retrieval of the summarized information from the documents collected from web based search engine as per the user query related to cricket and hockey domain. The system is designed in a manner that it takes the voice commands as keywords for search. The parts of speech in the query are extracted using the natural language extractor for English. Based on the keywords the search is categorized into 2 types: - 1.Concept wise - information retrieved to the query is retrieved based on the keywords and the concept words related to it. The retrieved information is summarized using the probabilistic approach and weighted means algorithm.2.Keyword search - extracts the result relevant to the query from the highly ranked document retrieved from the search by the search engine. The relevant search results are retrieved and then keywords are used for summarizing part. During summarization it follows the weighted and probabilistic approaches in order to identify the data comparable to the keywords extracted. The extracted information is then refined repeatedly through the aggregation process to reduce redundancy. Finally the resultant data is submitted to the user in the form of audio output.