Search Classifier for Better Discovery Experience

Some books or journals have been identified to be consistently omitted by our library’s bento box style search system. To increase its precision, a search format classifier was constructed using a machine learning python library. The classifier was trained with search history and associated click categories as correct answers (books, journals, databases, articles and websites). The idea is: If the returned probability that a search intends to find what belongs to a particular category is significantly higher than the other 4, the corresponding bento box will be expanded to include the desired item.

Beginner Poster
Location: Poster Area Date: November 11, 2017 Time: 3:15 pm - 4:15 pm Minhao Jiang