Towards optimize-ESA for text semantic similarity: A case study of biomedical text

Explicit Semantic Analysis (ESA) is an approach to measure the semantic relatedness between terms or documents based on similarities to documents of a references corpus usually Wikipedia. ESA usage has received tremendous attention in the field of natural language processing NLP and information retrieval. However, ESA utilizes a huge Wikipedia index matrix in its interpretation by multiplying a large matrix by a term vector to produce a high-dimensional vector. Consequently, the ESA process is too expensive in interpretation and similarity steps. Therefore, the efficiency of ESA will slow down because we lose a lot of time in unnecessary operations. This paper propose enhancements to ESA called optimize-ESA that reduce the dimension at the interpretation stage by computing the semantic similarity in a specific domain. The experimental results show clearly that our method correlates much better with human judgement than the full version ESA approach.

Enriching Topic Coherence on Reviews for Cross-Domain Recommendation

With the advent of e-commerce sites and social media, users express their preferences and tastes freely through user-generated content such as reviews and comments. In order to promote cross-selling, e-commerce sites such as eBay and Amazon regularly use such inputs from multiple domains and suggest items with which users may be interested. In this paper, we propose a topic coherence-based cross-domain recommender model. The core concept is to use topic modeling to extract topics from user-generated content such as reviews and combine them with reliable semantic coherence techniques to link different domains, using Wikipedia as a reference corpus. We experiment with different topic coherence methods such as pointwise mutual information (PMI) and explicit semantic analysis (ESA). Experimental results presented demonstrate that our approach, using PMI as topic coherence, yields 22.6% and using ESA yields 54.4% higher precision as compared with cross-domain recommender system based on semantic clustering.