deepSearch

Visit us at: wikideepsearch.com/

DeepSearch is a hybrid search solution that combines deep learning, vector space modeling and traditional search technology to create a unique enterprise and/or web search experience.

In deepSearch, queries are entered à la Google via a search box; however, by offering an autosuggest service, deepSearch behaves like an intelligent assistant that attempts in realtime to target lock the subject of the query.

Entire documents can also be copied and pasted into the search box before being compared to other documents for semantic similarity: results are sorted before being displayed by order of relevance.

By using a document as a query, deepSearch effectively becomes a concept search engine enabling, for example, academics, medical professionals, legal researchers or investment analysts to carry out a search with a research paper, a medical guideline, a court’s jurisprudence or a press article.

Finally, both keyword and document queries can be saved inside the user’s browser, thus saving time whilst providing anonymity to the user i.e. no information is stored on the server, deepSearch being a totally stateless - as in no session - web service.

Note

This Wikipedia search client is meant to showcase deepSearch’s unique technology offering and is not in any way a replacement for Wikipedia search.

Also, the delay incurred during the display of search results is due to deepSearch’s reliance on Wikipedia’s mediawiki servers for returning article snippets: the deepSearch service, which employs tiny document vectors (100 dimensions), only takes a few milliseconds to produce sorted results from millions of stored articles.