I can’t believe it is the end of 2016 already. Time is flying by. The Research Department led by Professor Candidate Ravi performed amazing research in 2016: Context sensitive search platform.
Context Sensitive Search Platform
The idea behind the development of a context sensitive search platform was to expanded and trained on various e-commerce and text-base web sites. Presently there is no search appliance that does not use statistical analysis in parsing search terms. What we propose in a novel system that can parse the context of it’s textual base (for instance, the bible) and answer context¹ sensitive queries such as “what is the salary of judges”. Existing methods seek to match the frequency of “judge” with “salary” and report results. What we are doing is to develop a context around “salary of judges” that understand that “Daniel” in the contextual search (the bible) is a judge and that it should look for “wages” as well as salaries.
Basic Level/Existing Domain Knowledge
We did due diligence technical research and have identified a number of technical gaps we need to overcome. Available industry learning is based on statistical analysis of text and machine learning algorithms. There is no available “plug-and-play” package that analyzes word hypernyms, hyponyms and sister terms to develop a richer framework in which to place a search term. There exists no online API that provides sentence frames for words. We were unable to locate any sort of software that would build a contextual web, a lexical directed graph, or even provide a complete list of example contexts that a word or search term could be found in. BTW, Google patent word analysis is “stem² based” only. See patent #US8990066, patent #US8527262 and patent #US7409334.
Uncertainties
To the best of our knowledge we are taking an original approach to contextualizing searches. What we propose is to augment existing search strategies with a contextual approach. Existing search contextualization uses machine learning in an orthodox way and we are using them in a novel way with a contextual base for the search space. What it means is we have to build our contextualization system and then seek how to apply machine learning algorithms, which partially we did in our OBS Site Search Product. There are further challenges with programming languages. parallel processing to build a rich contextualization in real time, operating systems, interaction between machine learning and AI algorithms³, and use case development methodologies.
Sources
- Google TrustRank https://www.search3w.com/google-new-algorithms-and-search-engines-changes/
- Read more about Stem & Morphological Analysis https://www.search3w.com/?s=stem
- AI https://www.search3w.com/googles-rankbrain-is-an-old-hat-artificial-intelligence-machine/
- Patent #US8990066 – Resolving out-of-vocabulary words during machine translation
- Patent #US8527262 – Systems and methods for automatic semantic role labeling of high morphological text for natural language processing applications
- Patent #US7409334 – Method of text processing