Many solutions are called semantic search engines when synonyms are used while searching. It is clear that systems should search for ‘software developer’ and ‘programmer’ and many more …. but that is only one aspect of semantic search! Software providers tend to confuse the market with terminology that doesn’t really match with what the product really offers!
Some systems are using ‘concepts’ – combinations of words such as ‘sales manager’, ‘business developer’,… – to search. But often the reality is that these systems simply execute a standard search on the different parts. Just search on ‘business developer’ and see how often ‘business intelligence developer’ is appearing in the results…. Not at all what you expect and totally wrong as these systems search for ‘business’ and ‘developer’ and ignore that the word ‘intelligence’ is adding a totally different context.
Actonomy is about the only robust solution that uses a combination of advanced semantic techniques that are required to get the label of ‘semantic search/match engine’ :
– searching on synonyms! Obviously we are using synonyms to search! Even more our xMP ontology has about the richest dictionary with synonyms in the industry – over 400,000 concepts and more than 1 billion words are recognized! Furthermore, the xMP ontology is constantly improving through several releases per year.
– searching on related concepts! Related concepts are concepts that are not a synonym but that are related even when the meaning is expressed in totally different words. Example : ‘business developer’ and ‘new market manager’ – totally different concepts but related. Thanks to the advanced machine learning systems of Actonomy, we are capable of knowing that these concepts (titles in this example) are related. Check out www.actonomy.com for more details on the advanced research projects we are conducting.
– interpretation of the context in which concepts are found! The context in which a concept is found is important. Example : ‘I was working with local sales managers and product managers.’ If this sentence is written in one’s CV, than it should not pop-up in the results while searching for ‘sales manager’. ‘Sales manager’ is in this context not relevant. On the other hand if in one’s CV it is written, ‘I was managing a sales team of 5 account manager’, it should be interpreted as ‘sales manager’ – the context of ‘managing’ and ‘sales team’ defines the job title ‘sales manager’.
– linguistic processing to handle typos and spelling! Our linguistic processing engine is handling singular and plural as well as feminine and masculine expressions, customized for each language.
– semantic matching requires another dimension of overall programmable profile matching. It is not only because job titles are found as synonym or related concept that there is a good match. There is so many different categories of criteria such as skills, education, implicit knowledge, domain experience and many more that can make that a match is good or bad!
Our ongoing research in ontology modeling, machine learning and artificial intelligence has resulted in advanced solutions that really do semantic search and semantic matching! A profile match integrates all the complexity of semantic search and job titles, skills, etc. and creates a total view on the one’s profile.