BOTTARI: Location based Social Media Analysis with Semantic Web

2,053 views
1,919 views

Published on

Bottari is a LarKC application http://www.larkc.eu/. It offers a real-time personalized recommendation service for restaurants in Insa-dong(Seoul) listening to the reputation of the restaurants on social media. Social media anlytics is powered by LarKC inductive and deductive stream reasoning solution. Learn more at http://larkc.cefriel.it/lbsma/bottari/ .

Published in: Education, Business
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
2,053
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
56
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide
  • Service-Finder
  • BOTTARI: Location based Social Media Analysis with Semantic Web

    1. 1. BOTTARI: Location based Social Media Analysis with Semantic Web Emanuele Della Valle Joint work with: CEFRIEL : Irene Celino, Daniele Dell ’ Aglio, Marco Balduini SALTLUX : Tony Lee, Seonho Kim S IEMENS : Volker Tresp, Yi Huang
    2. 2. Watch this first :-) 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany http://www.youtube.com/watch?v=c1FmZUz5BOo
    3. 3. <ul><li>An augmented reality application for personalized recommendation of restaurants in Seoul </li></ul>What have you seen? 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    4. 4. <ul><li>Yes and no! </li></ul><ul><li>Same use case, more “democratic” </li></ul><ul><li>We do “reality mining” by listening to the social media </li></ul>Yet another ? 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    5. 5. Architecture 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany out Query Rewriter Query Evaluator RDF2Matrix Plug-in Streaming Linked Data Server SOR Invoker SOR geo-spatial KB Social Media Crawler and Sentiment Miner HTTP PULL: Query Initiated PUSH: Data Initiated SPARQL androjena
    6. 6. Sentiment Mining <ul><li>Precision tests: </li></ul><ul><ul><li>Auto-generated rules ≈ 70% </li></ul></ul><ul><ul><li>Manually-coded rules ≈ 90% </li></ul></ul><ul><ul><li>Syllable kernel ≈ 50~60% </li></ul></ul><ul><li>Our target > 85% </li></ul>26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany Micropost message Morphologically Analyzable? Rule based Analysis Auto generated rules Learned documents SVMs Syllable Kernel Sentiment of the tweet Yes No
    7. 7. SOR - Geo-Spatial KB 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    8. 8. C-SPARQL and Streaming Linked Data Server 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    9. 9. <ul><li>A machine learning framework for inductive materialization </li></ul><ul><ul><li>Detects interesting data patterns </li></ul></ul><ul><ul><li>Predics RDF-triples </li></ul></ul><ul><ul><ul><li>i.e., which restaurant a user will tweet positively about </li></ul></ul></ul><ul><li>Caractheristics </li></ul><ul><ul><li>Capability to deal with sparse, high-dimensional and incomplete data </li></ul></ul><ul><ul><li>Multivariate latent space based approach </li></ul></ul><ul><ul><li>Modularized approach for easily integrating contextual information </li></ul></ul>SUNS (Statistical Unit Node Sets) 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    10. 10. <ul><li>SELECT DISTINCT ?poi ?name ?lat ?long ?numPos ?prob </li></ul><ul><li>WHERE { </li></ul><ul><li>?poi a ns:NamedPlace ; </li></ul><ul><li>ns:name ?name ; </li></ul><ul><li>geo:lat ?lat ; </li></ul><ul><li>geo:long ?long . </li></ul><ul><li>FILTER (f:within_distance( 37.5 , 126.9 , ?lat, ?long, 200 )) </li></ul><ul><li>FILTER (f:dest_point_viewing( 37.5 , 126.9 , ?lat, ?long, 90 , 200 )) </li></ul><ul><li>{ :someUser sioc:creator_of ?tweet . </li></ul><ul><li>?tweet twd:talksAboutPositively ?poi . </li></ul><ul><li>WITH PROBABILITY ?prob </li></ul><ul><li>ENSURE PROBABILITY [0.5..1) } </li></ul><ul><li>?poi twd:numberOfPositiveTweets ?numPos . </li></ul><ul><li>} </li></ul><ul><li>ORDER BY DESC(?numPos), ?prob, f:distance( 37.5 , 126.9 , ?lat, ?long) </li></ul><ul><li>LIMIT 10 </li></ul>Query Processing 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany GEO-SPATIAL PROBABILISTIC STREAMING
    11. 11. LarKC At Work 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany PULL: Query Initiated PUSH: Data Initiated SPARQL androjena Probabilistic part of the query to get personalized recommendations (the “ for me ” button in BOTTARI) Geo-Spatial part of the query to get POIs closer to user location Streaming part of the query to get trends in users' sentiment (the “ emerging ” button in BOTTARI) Input user query is split Results of the different computations are joined out Query Rewriter Query Evaluator RDF2Matrix Plug-in Streaming Linked Data Server SOR Invoker SOR geo-spatial KB Social Media Crawler and Sentiment Miner HTTP
    12. 12. Evaluation - Efficacy 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany 5 10 15 20 25 30 0,7 random knnItem emerging (C-SPARQL) for me (SUNS) SUNS + C-SPARQL 0,6 0,5 0,4 0,3 0,2 0,1
    13. 13. Evaluation - Efficiency 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany Hardware: 2.66 GHz Intel Core 2 Duo with 8 GB RAM
    14. 14. Evaluation – Scalability 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany Number of concurrent users Query Latency (sec)
    15. 15. <ul><li>End-user application </li></ul><ul><li>Attractive and functional interface </li></ul><ul><li>Real-world dynamic data </li></ul><ul><li>Fully based on Semantic Web technologogies </li></ul><ul><ul><li>RDF as common data format between heterogenous components </li></ul></ul><ul><ul><li>SPARQL as query language </li></ul></ul><ul><li>Rigorously evaluated </li></ul><ul><ul><li>Effective </li></ul></ul><ul><ul><li>High throughput for handling dynamic data </li></ul></ul><ul><ul><li>Scalable in number of concurrent users </li></ul></ul><ul><li>Commercial Potential </li></ul>Conclusions 26.10.2011 - SW Challenge 2011, ISWC 2011, Bonn, Germany
    16. 16. Emanuele Della Valle Joint work with: CEFRIEL : Irene Celino, Daniele Dell ’ Aglio, Marco Balduini SALTLUX : Tony Lee, Seonho Kim S IEMENS : Volker Tresp, Yi Huang Any question?

    ×