Συλλογές
Τίτλος Creating a repository for data crawled from multiple social networks
Δημιουργός Christofilos, Konstantinos
Συντελεστής Athens University of Economics and Business, Department of Informatics
Kotidis, Yannis
Τύπος Text
Φυσική περιγραφή 37p.
Γλώσσα en
Περίληψη People use countless web services in their everyday life. The same thing happens to companies and organizations. Most cloud services give access to the generated data via REST APIs, but although this seems very nice, it becomes a headache when someone wants to get results from more than one service.The main problem is that API responses are not compatible among different services and the solution to that is the transformation of these semi-structured data to structured data.The second problem, and probably the most hard to tackle, is to find relations in these data and produce some worthy results.The approach that was adopted on this thesis was to fetch data from two input API sources (Twitter, Instagram), find relations between the data using natural language processing in the responses and finally merge them into a structured data environment. In that case, we manage to have a single query interface on the raw data and in addition, some relations pre-populated.The merged output will be stored in two different graph databases, one property graph (Neo4J) and one semantic graph (RDF-Apache Jena). In that way we had the ability to give a unified query environment by using one of each databases.
Λέξη κλειδί Named Entity Recognition (NER)
Representational State Transfer (REST)
Application Programming Interface (API)
Resource Description Framework (RDF)
Natural Language Processing (NLP)
Ημερομηνία 31-10-2016
Άδεια χρήσης https://creativecommons.org/licenses/by/4.0/