Helping users sort faster with adaptive machine learning recommendations

摘要: Sorting and clustering large numbers of documents can be an overwhelming task: manual solutions tend to slow, while machine learning systems often present results that don't align well with users' intents. We created evaluated a system for helping users sort into clusters. iCluster has the capability recommend new items existing clusters appropriate items. The recommendations are based on model adapts over time - as user adds more cluster, system's improves become relevant. Thirty-two subjects used hundreds data both without recommendations; we found allow rapidly. A pool 161 raters then assessed quality resulting clusters, finding generated were statistically indistinguishable quality. Both assisted methods substantially better than fully automatic method.