Sentēme

About

Learn from your fellow consumers which products to buy and which to skip.

Sentēme (pronounced sen-teem) works with a suite of linguistic classifiers and machine learning algorithms to intelligently harvest, parse, analyze, and summarize unstructured opinion data associated with products with (semi-)structured features on the web.

We use a hybrid approach to information retrieval and sentiment mining in which a rules-based NLP approach is being used as a pre-processing step to translate unstructured text into semi-structured data which can then be leveraged within a machine learning model designed to simulate human interpretation of textual data (and by proxy human assessments of products); to get at substantively finer-tuned automated analysis of human opinion data. The Sentēme Engine goes beyond the use of term vocabularies or raw text algorithms by leveraging predictive power of supervised learning algorithms in tandem with robust linguistic classifiers that encode syntactic, semantic and pragmatic feature sets.