LexExp

LexExp expands lexicons by generating spelling variants and multi-word lexical expressions to improve analysis of noisy biomedical texts.

Key Features:

• Phrase Embedding Model: Employs a phrase embedding model that captures semantic relationships between words to generate contextually relevant lexical variants.
• Lexical Similarity-Based NLP Methods: Uses lexical similarity measures and NLP techniques to identify and expand terms within biomedical texts.
• Tunable Threshold Decay Functions: Incorporates tunable threshold decay functions to balance recall and precision during variant generation.
• Multi-Word Expression Handling: Handles multi-word expressions to expand complex biomedical terminologies.
• Spelling Variant Generation: Generates spelling variants to address noise and orthographic variation in biomedical text.

Scientific Applications:

• Text Mining: Enhances extraction of biomedical information by adding spelling and lexical variants to lexicons.
• Data Integration: Facilitates standardization of terminology across datasets by expanding lexicons to include variant spellings and expressions.
• Semantic Analysis: Improves semantic analyses for tasks such as gene-disease association studies and drug repurposing by broadening term coverage.

Methodology:

LexExp generates spelling variants and expands lexical expressions using phrase embeddings and lexical similarity measures, with tunable threshold decay functions to adjust recall versus precision and explicit support for multi-word expressions.