Math-Similarity Search (MSS)

The unique structural syntax and the variety of semantic equivalences of mathematical expressions make it a challenge for a keyword-based text search engine to effectively meet the users' search needs. Many existing math search solutions focus on exact search where the notational matching determines the relevance rank, while the structural similarity and mathematical semantics are often missed out or not addressed adequately. One important research question is how to effectively and efficiently find math expressions that are similar to a user's query, and how to do relevance ranking of hits by similarity. This research focuses on (1) conceptualizing similarity between mathematical expressions, (2) defining metrics to measure math similarity, (3) utilizing those metrics for math-similarity search (MSS), (4) optimizing the proposed MSS systematically, and (5) evaluating performance to validate advantage of the MSS. The findings of this research show that the performance of the proposed math similarity search is superior to that of keyword based math search with respect to both relevance ranking and recall.

Author

• Zhang, Qun

Language

• en

Date created

• 2015

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Computer Science

Degree

• Ph.D.

Advisor

• Youssef, Abdou S

Committee Member(s)

• Zhang, Nan
• Miller, Bruce R
• Choi, Hyeong-Ah

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/dr26xx473