Read out

Guest Talk: On Unified Stream Reasoning

Daniele Dell'Aglio

Date/Time: Thursday, 18.02.2016, 11:00

Loca­tion: D2.2.094


The real-­time inte­gra­tion of huge volumes of dynamic data from hete­ro­ge­neous sources is getting more and more atten­tion, as the number of data­-stream sources is keeping growing and chan­ging at very high pace. While Data Stream and Event Proces­sing deal with data streams and reac­tiveness, Reaso­ning is a poten­tial solu­tion for the data hete­ro­gen­eity: onto­lo­gies enable access to data streams from diffe­rent sources and make explicit hidden infor­ma­tion. Stream Reaso­ning aims at brin­ging toge­ther those areas, with tech­ni­ques to perform reaso­ning tasks over data streams. In this context, the problem I inves­ti­gate is how to unify the current Stream Reaso­ning tech­ni­ques, as they may substan­ti­ally differ from each other. This fact is evident when these tech­ni­ques are desi­gned to reach diffe­rent goals, e.g. aggre­ga­ting data in the stream vs. detec­ting events. However, it happens even when they perform the same task and final users may expect the same beha­viour. Under­stan­ding pecu­li­a­ri­ties and common points is manda­tory in order to compare, contrast and inte­grate them.The main outcome of this rese­arch is RSEP-QL, a formal model to describe the evalua­tion seman­tics of stream reaso­ning systems in the context of conti­nuous query answe­ring. RSEP-QL extends SPARQL by adding opera­tors to manage streams such as sliding windows and event patterns. Simi­larly to SPARQL, RSEP-QL works under entail­ment regimes, which intro­duce deduc­tive infe­rence in the conti­nuous query answe­ring process. The value of RSEP-QL is shown through two appli­ca­tions in the areas of compa­ra­tive testing and query opti­miza­tion.


Daniele Dell'Aglio is a PhD student at the Dipar­ti­mento di Elett­ro­nica, Infor­ma­zione e Bioin­geg­neria (DEIB) of the Poli­tec­nico di Milano since November 2012. He is advised by Prof. Emanuele Della Valle, and his rese­arch activity focuses on Stream Reaso­ning, i.e. the appli­ca­tion of infe­rence tech­ni­ques to data streams. In his major rese­arch topic, Daniele studies the problem of unifying Stream Reaso­ning tech­ni­ques in the context of conti­nuous query answe­ring. Daniele won an IBM PhD Fellowship award 2014, and he is currently involved in the activi­ties of the W3C Commu­nity Group on RDF Stream Proces­sing. From 2008 to 2012, Daniele worked as junior rese­ar­cher and consul­tant at CEFRIEL. He parti­ci­pated in the Smart City rese­arch activi­ties of the LarKC FP7 project and in rese­arch activi­ties related to Web services and recom­mender systems in the SOA4All and the Service Finder FP7 projects. Daniele holds a MSc and a BSc in compu­ting system engi­nee­ring (Poli­tec­nico di Milano). He cont­ri­buted in the realiza­tion of several proto­types of services in the urban context, such as BOTTARI (1st prize at the Semantic Web chal­lenge 2011), Traffic LarKC (1st prize at the AI Mashup chal­lenge 2011), Twindex and ECSTASYS (respec­tively 3rd prize at the AI Mashup chal­lenge 2013 and 2014).

Back to overview