Enhancing Query Expansion Method Using Word Embedding

Enhancing information retrieval with semantic query expansion: a Word2Vec-based approach

“…Where λ is the tuning parameter, R(Q, d) represents the relevance between the original query Q and the document d, and R(Q_exp, d) denotes the relevance between the expanded word set and the document d. The calculation formula is as Equation ( 6) and (7).…”

“…Yusuf presents an improved query expansion method by combining the unigram model with Okapi BM25 and word embeddings using Glove. Experimental results on the Arberry dataset show that the proposed method using Glove word embeddings significantly enhances query expansion methods, addressing the limitations of previous search applications [7] . Rosin proposes a method to improve retrieval performance in event-related queries by leveraging events and utilizing a novel mechanism for query expansion, which significantly outperforms stateof-the-art methods on newswire TREC datasets [8] .…”

Warrant Generation through Deep Learning

“…[24] suggested Causality-attention: A convolutional neural network with multiple columns for why-QA. The claim expansion process in our work is inspired by (question query Q) [25]- [27], which employs a word embedding to expand the query (in our work, claim) and wordnet expansion [28]. The model checks for hypernyms, such as food, and hyponyms, such as fruit, in addition to meronyms and holonyms; a branch is a meronym (part meronym) of a tree, whereas heartwood is a meronym (substance meronym) of a tree, and the forest is a holonym (member holonym) of a tree.…”