ASIMOV aims to improve component technologies associated with semantic job matching engines so that the right candidates can be automatically matched to job openings. To that end, ASIMOV will combine big data analytics with natural language processing and machine learning to derive ontologies from multilingual text data automatically. The project will also explore the semantic interpretation of additional user profile-related information, such as browsing patterns.
Using language to define meaning
The market for advanced job-matching software is enormous. Furthermore, most job markets, and European ones in particular, are multilingual by nature. Current engines for job matching rely heavily on manually-created and thus expensive semantic cores to understand unstructured job-related information. As a result, it is often difficult and time-consuming to adapt this core to a new domain or language.
The ASIMOV project explores how semantic matching engines can learn from unstructured multilingual textual data instead of relying on human-created and translated ontologies. This should allow the fast creation of new ontologies and reduce the maintenance effort of existing ones, leading to more up-to-date, accurate information matching in more languages.
Combining semantics with user profile data
Current matching engines in the human resource domain are also inadequate in using contextual information such as user profile information and customer behaviour in the matching process. ASIMOV will explore techniques to extract semantic information from such contextual information in order to further enhance the performance of the new matching engine. The use of contextual information should not only lead to more relevant results when searching for jobs, but should also result in better-targeted job-related ads.
Lower costs, more languages, further reach
The ASIMOV consortium partners expect the maintenance time of semantic matching systems, SEO costs and the development time for new languages to be significantly reduced. This not only opens up opportunities in new language regions but also allows rapid expansion into new domains and applications. In the human resources domain, the new technology will give vacancies a greater reach and increase job view traffic. This should help companies fill job openings more quickly, resulting in an overall boost to the economy. The new techniques are also intended to bring the job-seeking process closer to the user, reducing unemployment and increasing mobility.
The ASIMOV project will contribute to decreasing development costs and time of semantic matching systems and increase matching quality, providing industry partners with the capabilities required to consolidate and expand their markets.
Multilingual & profile-based semantic matching for human resources applications.
ASIMOV is an imec.icon project funded by imec and IWT
It ran from 01.04.2016 until 30.06.2018.
- De Persgroep
- imec - Data Science Lab - UGent
- imec - ITEC - KU Leuven
- Project Lead: Luc Mertens
- Research Lead: Kris Demuynck
- Innovation Manager: Dirk Hamelinck