Customer problem reports, beyond ‘how do I’s and ‘feature asks’ tend to vary widely from being cryptic to long and explanatory. Summarizer solves challenges associated with understanding core of problem from long statements and attachments.
Sainapse
- Identifies sentences in a text document and maps each sentence into high dimension vector space
- Buckets sentences into ‘concept clusters’ and recognizes dominant ‘concepts’ that’s articulated in the problem report. These ‘concept clusters’ form core of Sainapse’ way of identifying real issue often hidden in complex sentences.
- Based on these concepts then builds ‘concept relationship’ that defines the ‘context’ of reported problem
- Identifies and moderates sentences occurring across ‘concept clusters’ via regularization techniques to reflect its appropriate weightage
- Finally extracts top 2-3 sentences from long descriptions and attachments as summary
Sainapse summarizes knowledge documents when it build the mental model at the time of training. As for real time cases arriving with long texts and attachments Sainapse summarizer works in the way described above to zero in, on the problem and then triggers Recommender to take over from there.
Summarizer - So that you can select before you double click
