Trovix’s unique fuzzy logic search functionality is saving our recruiting staff huge amounts of time in identifying the most qualified resumes for our hundreds of job openings. We’re finally able to take advantage of the thousands of resumes in our database. Trovix has also provided tremendous service and support to make sure our staffing organization can meet its goals. 
Trovix human-like search technology is a breakthrough that combines complex scoring algorithms, learning, and a comprehensive knowledge base to provide personalized search results for recruiters and job seekers. Trovix understands the true context of the information on a resume and included in a job description and instantly identifies the best qualified candidates for that specific position in a way that no other search technology can.
Traditional Approaches
Searching full-text documents (like resumes or job requisitions) with the search technology built into most job boards or Applicant Tracking Systems today can lead to erroneous and sub-optimal results. On a basic level, most search engines will not recognize misspelled words, abbreviations, concatenated words, missing spaces or varying word order. In addition, most search engines will yield redundant and non-related results for polysemous words, or words that have multiple meanings like “Java.” Perhaps most importantly, unless a search engine understands the true hierarchical relationships between universes of concepts (i.e. understanding that the “Haas School of Business” is a part of University of California Berkeley), highly relevant documents will be missed altogether. In the context of recruiting, the challenge for search technologies is further complicated because users often do not know exactly what they are looking for until outcomes are experienced. A hiring manager may not be able to express exactly what he is looking for until after he sees a few resumes and can refine his search. In addition, two hiring managers may rank the same set of candidates completely differently for the same position, based on their preferences. Recruiting is very personalized and the search technology must accommodate that.
The Trovix Approach
The Trovix search engine addresses all of the above problems by using a set of tightly integrated technologies that combine automatic concept extraction/matching from text, a powerful fuzzy search engine, and a collaborative user preference learning engine to provide highly accurate and personalized search results. It normalizes textual input into a meta-representation (according to a set of rules) consisting of concepts and relationships between concepts, into a proprietary knowledge base. It then normalizes the search query into the same meta representation to find matching items.
In other words, first a concept extraction and classification engine breaks the document down into key concepts and the relationships that exist between these concepts. These are understood by Trovix's hierarchical (aka subsumptive) knowledge base. Taking the U.S. geo-political structure as an example, cities are grouped into counties, counties into states, states into countries, and so forth. Further, telephone area codes are linked to cities, town mayors are linked to cities and local grocery stores are linked to cities. If a query is performed for a town's mayor, the search engine should automatically understand that it is related to a specific city. With resumes, a person's set of skills should be understood even if different words are used to describe it. If a search is conducted for “Berkeley” for example, all subordinate concepts should be dynamically identified. This is one of the many differentiating factors of Trovix's breakthrough search technology.
When a job requisition is entered into the system, it too is automatically distilled into its meta representation and a Degree of Match is calculated between it and each potential resume. The calculations, conducted on a concept by concept basis, determine how closely a candidate's experience matches the job requirements. Another unique differentiating capability of Trovix's search technology is its embedded Profiled Search Engine that personalizes search results for each user in the context of a search. It integrates a learning algorithm that automatically determines the value of certain criteria relative to other criteria. This patented method of collaboratively building user preference profiles is applied in the search algorithms for each individual concept.
