Indian Journal of Science and Technology

Abstract

Objective: Computing semantic similarity is a significant objective in Information Retrieval (IR) due to the emerging web searching process and the demand of fast and efficient result provisioning. Methods: Exploiting static lexical resources leads most of the conventional semantic similarity measurement techniques to ignore temporal aspects of concepts. Later, to resolve the temporal constraint, the research works exploit independent sources like Newspaper articles, television documentaries, and books. However, these methods meet the data-sparsity problem, since the independent sources are too restrictive and exclude articles that are worthy of notice. Findings: In order to provide the precise result for recent trend concepts, this paper utilizes the social media as the robust and dynamic source, often reflect public opinion. This work introduces an ideNtifying nEw concept Evolution thRough semantIC Social media temPoral analYsis (NEOTERICSPY) model incorporating two phases, namely semantic relatedness measurement in a time series and examining temporal fluctuations between time series. Initially, the NEOTERIC-SPY measures the semantic relatedness of a word-pair using normalized distance measurement in a time series and facilitates the identification of temporal fluctuations. It reduces the word-pair list of a concept among the overall possible word-pairs and provides a good trade-off between the execution time and accuracy. Secondly, it identifies the new concepts using temporal correlation analysis of the semantic similarity measurement between the time series. It employs the weighing function of the temporal dynamics to find the recent changes than the ancient changes of the concepts. The experimental results show that the NEOTERIC-SPY approach significantly outperforms the conventional method and illustrates the consistent improvements of its performance. Application/ Improvements: The NEOTERIC-SPY leads the IR system to provide the desired information to the users in a short-time period, ensuring the precise result pertaining to the recent concept.

Keywords: Semantic Similarity, Social Media, Temporal