Retrieval of Mathematical Information with Syntactic and Semantic Structure over Web

摘要

Efficient retrieval of mathematical expressions over web is a complex process as compared to simple text search. This is only possible when the syntactic (e.g. Textual) and semantic (e.g. Structural) information of a mathematical expression is retrieved properly and analyzed methodically. In this paper, we are proposing a technique that indexes expressions along with their syntactic and semantic information. These expressions are represented in ContentMathML(CMML). To improve the memory efficiency in index, an encoding technique is introduced which encode CMML mathematical expressions in Braille Unicode characters. In order to improve ranking of retrieved documents, a weighting function is introduced which assign a weight to each indexing term. The weighting score of each term contributes in ranking function that improves the rank of a document which contains query terms. The proposed technique is evaluated on NTCIR-12 Wikipedia and Arxiv corpora. Performance is also measured using NTCIR-MathIR evaluation criteria. The precision for Wikipedia-formula-queries is achieved 47% and for Arxiv is achieved 44% at top 5 documents.