Designed and developed relation embedding based supervised classification method to detect synonym pair hidden in Japanese query log where distributed embedding feature is used to measure the semantic similarity, then string and morphological features are used to catch expression similarity and syntax similarity
Designed and implemented a query-embedding based semantic model to making suggestion list for those queries which are lying in tail of query log and cannot be used to make suggestion list by frequency term.
2015-10 - 2016-4
Automatic address correction for incorrect user input in Credit-card system
- Designed and developed a tool for correcting English address spelling and sequence error for Credit-card application and account management system. - Develop a model for English segmentation by using naïve-Bayes.
2014-12 - 2015-10
Shopping commodity name recognition
Developed a CRF-based shopping commodity name recognition tool from E-commerce user review.
2014-12 - 2015-8
Rephrasing Adjective Containing Queries on E-commerce Sites
Designed and implemented a model to analysis query semantic from user shopping intent by user reviews.