A team of Aalto won the TekesHack Design Contest with their semantic engine
Semantic matching algorithm can be utilized to find related documents in a wide range of applications.
The challenge of TekesHack Design Contest was to combine open data, machine learning and artificial intelligence in order to create a whole new service solution and to demonstrate efficient ways to match companies, research teams and investors.
”This is the first time when hackathon is combined with a public procurement in Finland”, says Procurement Manager Eriika Autio from Tekes.
The winning team of the demo competition was Aalto University team with Tapio Auvinen from the Department of Computer Science and Vincent Kuo from the Department of Civil Engineering. Their excellent demo convinced the jury and deserved the first prize of 10.000 euros.
“We addressed the problem of knowledge tacitness, difficulty of matching different words and sentences, using data mining and machine learning algorithms to vectorise texts and concepts from relevant data repositories. This allows us to quantitatively gauge semantic similarities between texts using scalable computations, creating links between texts passages that are syntactically different, but are similar in meaning”, tells Vincent Kuo.
The prototype allows users to search for investors, startups, organizations/companies and researchers using keywords, sentences or longer texts such as research abstracts as the query.
The prototype mined data from Slush, Aalto and Tekes databases and enables the user to make queries using natural language. The researchers further provided a match-making and semantic mapping function in the tool for the user to build respectively relevant value-chains and sound collaboration, flexible enough for any actor of the innovation ecosystem – be it researchers, startups, large companies, public sector or investors.
“Matches are based on the meanings of the words and the query can be anything, a sentence or even a whole article. The semantic engine can do even cross-language queries, where utilizing for instance specific terminology of any one area in both languages. The semantic engine can be used for instance in looking for semantically related research data, competitors and even interesting jobs or employers”, concludes Tapio Auvinen.
Hackathon event was one phase of the Design Contest. Next phase of the Design Contest is to pass competitors’ work to an independent jury for an evaluation. The winner or winners of the final evaluation will make a contract with Tekes. The value of the deal varies between 100.000 and 200.000 euros.
Department of Computer Science
+358 40 756 2742
Department of Civil Engineering