“Our experiments showed that most open source stemmers resulted in either too many bad results from over-stemming (low precision) or too many missing results (low recall) compared to a lemmatizer. Rosette allowed us to optimize the balance between precision and recall to provide a superior experience for our customers’ searches.”

“We were able to get our internal assessment of search quality to a point where we felt comfortable shipping the product. Metrics measure how users engage with search results, and from human evaluation, the improvement is due to fewer misinterpreted search queries.”