Ce diaporama a bien été signalé.
Le téléchargement de votre SlideShare est en cours. ×

Discovering the Dynamics of Terms’ Semantic Relatedness through Twitter

Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Prochain SlideShare
resume_SDE
resume_SDE
Chargement dans…3
×

Consultez-les par la suite

1 sur 12 Publicité

Plus De Contenu Connexe

Similaire à Discovering the Dynamics of Terms’ Semantic Relatedness through Twitter (20)

Publicité

Plus récents (20)

Discovering the Dynamics of Terms’ Semantic Relatedness through Twitter

  1. 1. Discovering the Dynamics of Terms’ Semantic Relatedness through Twitter<br />Nikola Milikic, University of Belgrade, Serbia<br />JelenaJovanovic, University of Belgrade, Serbia <br />Milan Stankovic, STIH, Université Paris-Sorbonne, France<br />
  2. 2. Outline<br />Introduction<br />Normalized Micropost Distance<br />Scenarios of Use<br />Example Diagrams – Tweet Dynamics<br />Future Work<br />
  3. 3. Introduction<br />Micropost = Description of a Moment<br />
  4. 4. Introduction<br />Semantic Relatedness (SR) of terms is also a subject to temporal changes<br />Mutual relationship change of terms is not directly evident from simple query results<br />
  5. 5. Normalized Micropost Distance <br />Normalized Micropost Distance (NMD) - semantic similarity measure derived from the number of microposts containing a given set of keywords<br />Inspired by the Normalized Google Distance (NGD)<br />Google search results vs.Microposts<br />
  6. 6. Normalized Micropost Distance <br />NMD formula<br />x, y – terms<br />f(x)t , f(y)t – number of microposts for x and y<br />f(x, y)t– number of microposts containing both x and y<br />t – time interval<br />
  7. 7. Normalized Micropost Distance <br />Detecting the significance of change - standard deviation of NMDs<br />
  8. 8. Scenarios of Use<br />Adapting Online Advertising Campaigns to the Changes in Term Relatedness<br />Example: ‘sxsw’ and ‘ipad’<br />Facilitating Discovery of Relevant Resources in Organizations<br />harmonizing the official and the actual vocabularies within an organization<br />
  9. 9. Example Diagrams<br />Tweet Dynamics – demo application<br />NMD diagram for terms 'ipad' and 'sxsw' for the 5 days period<br />
  10. 10. Example Diagrams<br />Tweet Dynamics – demo application<br />NMD diagram for terms ‘japan' and ‘nuclear' for the 5 days period<br />
  11. 11. Future Work<br />work in progress<br />detection of good candidate term pairs<br />computational efficiency and the limits of Twitter API <br />comprehensive evaluation<br />test on the mass amount of data<br />compare to other approaches<br />Google Correlate<br />
  12. 12. Questions?<br />Nikola Milikic<br /> @milikicn<br /> http://nikola.milikic.info<br />

×