Invented a ranking system that re-ranks the search query results based on technical complexity of the documents.
- Predictive model uses a topological graph properties extracted from graph representation of text as input feature set.
- Graph representation captures conceptual cohesion, conveying author’s intent.
- Topological features captures cohesiveness, topic complexity and connectedness in constructing technicality score predictor.
- Attained predicting accuracy of more than 96% over test queries results, indicating high reliability.
- Patent pending for the above invention.