Related Experiment Videos
An implementation of the trigram phrase matching method for text similarity problems
Adalberto O Tardelli1, Meide S Anção, Abel L Packer
1Latin American and Caribbean Center for Health Science Information (BIREME/PAHO/WHO), São Paulo, Brazil.
Studies in Health Technology and Informatics
|March 8, 2005
Summary
This study introduces a trigram matching method using Term Frequency-Inverse Document Frequency (TFIDF) for efficient text similarity analysis. This approach enhances document retrieval and concept identification in scientific databases.
Area of Science:
- Information Science
- Computational Linguistics
- Bibliometrics
Background:
- Text representation using term vectors with TFIDF is effective for similarity problems.
- Identifying MeSH concepts, harmonizing citations, and normalizing affiliations are key challenges in scientific literature management.
Purpose of the Study:
- To implement and evaluate a trigram matching method for text representation and similarity.
- To improve the efficiency and feasibility of information services for scientific production discovery.
Main Methods:
- Utilized trigrams (overlapping 3-character strings) as terms in a text vector.
- Applied logarithmic Term Frequency-Inverse Document Frequency (TFIDF) for term weighting.
- Developed an efficiency-focused C-programming implementation with heuristic rules.
Main Results:
- The trigram matching method demonstrated effectiveness in text similarity tasks.
- The implemented algorithm, optimized for efficiency, proved feasible for information services.
- The approach aids in identifying similar texts and potentially synonym phrases.
Conclusions:
- Trigram-based TFIDF vector representation is a viable method for text similarity.
- The developed software tool and optimized algorithm enhance scientific information retrieval.
- This method supports services like discovering scientific production within collections such as SciELO.