1 / 18

Towards the Semantic Web: Collaborative Tag Suggestions (cited by 74)

Towards the Semantic Web: Collaborative Tag Suggestions (cited by 74). Zhichen Xu, Yun Fu, Jianchang Mao, and Difu Su Yahoo! Inc 2821 Mission College Blvd., Santa Clara, CA 95054 {zhichen, yfu, jmao, difu}@yahoo-inc.com. Workshop on Collaborative Web Tagging, WWW2006.

ronia
Télécharger la présentation

Towards the Semantic Web: Collaborative Tag Suggestions (cited by 74)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Towards the Semantic Web: Collaborative Tag Suggestions (cited by 74) Zhichen Xu, Yun Fu, Jianchang Mao, and Difu Su Yahoo! Inc 2821 Mission College Blvd., Santa Clara, CA 95054 {zhichen, yfu, jmao, difu}@yahoo-inc.com Workshop on Collaborative Web Tagging, WWW2006 Presented by Jun-Ming Chen 11/28/2008

  2. Outline • Introduction • Collaborative tag suggestion • A taxonomy of tags • Criteria for good tags • Collaborative Tag Suggestions • Tag Spam Elimination • Content-based Tag Suggestions • Tag Normalization • Temporal Tags • Adjustments • Examples • Conclusions

  3. Introduction • We propose a collaborative tag suggestion algorithm using these criteria to spot high-quality tags. • The proposed algorithm employs a goodness measure for tags derived from collective user authorities to combat spam. • The goodness measure is iteratively adjusted by a reward-penalty algorithm, which also incorporates other sources of tags, e.g., content-based auto-generated tags. • Our experiments based on My Web 2.0 show that the algorithm is effective.

  4. MyWeb 2.0

  5. Introduction

  6. Collaborative tag suggestionA taxonomy of tags • Content-based tags • Autos, Honda Odyssey, batman, open source • Context-based tags • tags describing locations and time such as San Francisco, Golden Gate Bridge, 2005-10-19 • Attribute tags • author of a piece of content such as Jeremy’s Blog and Clay Shirky • Subjective tags • Tags that express user’s opinion and emotion such as funny or cool • Organizational tags • my paper or my work • Golder and Huberman have also discussed tag categorization [3] [3] Golder, Scott A., Huberman, Bernardo A. “The Structure of Collaborative Tagging Systems.” HPL Technical Report. 2005.

  7. Collaborative tag suggestionCriteria for good tags An object, a specific tag, a generic tags • High coverage of multiple facets • High popularity (IR: term frequency) • Least-effort (number of tags :minimize ; number of objects: small) • Uniformity (normalization) (syntactic variance, synonym) • Exclusion of certain types of tags (matching the prefixes of the tags entered by the user before)

  8. Collaborative tag suggestionCriteria for good tags

  9. Collaborative tag suggestionCollaborative Tag Suggestions & Tag Spam Elimination • Ps (ti | tj ; o) • Pa (ti | tj) (any user) object o(the same user) ti tj (Reward) the overlap in terms of the concepts between ti and tj minimize the overlap of the concepts identified by the suggested tags

  10. Collaborative tag suggestionCollaborative Tag Suggestions & Tag Spam Elimination • S (t, o) • a (u) Goodness measure (score) of the tag t to an object o the goodness measure of a (tag, object) pair is the sum of the authority scores of all users who have tagged the object with the tag In asimple case where we assign uniform authority score of1.0 for every user. user (t, o) denotes the set of users who have tagged a given object o with the tag t. User u object tag S( t ,o ) apple pc 1.8 machine 0.6 red 0.8 car autos 2.6 BMW 1.3 machine 0.4 the authority score of a given user u the average quality of a given user’s tags object (u) is the set of objects tagged by the user u tag (o, u) denotes the set of tags assigned to object o by user u

  11. Collaborative tag suggestionCollaborative Tag Suggestions & Tag Spam Elimination • At each step, after a tag ti is selected, we adjust the score for each remaining tag t’ as follows: • Penalize tag t’ by removing the redundant information, • e.g., by subtracting Pa (t’ | ti) * S (ti ,o) from S (t’, o), • i.e., • This minimizes the overlap of the concepts identified by the suggested tags. • Reward tag t’ if it co-occurs with the selected tag tiwhen users tag object o. This rewarding mechanism also improves the uniformity of the suggested tags

  12. Collaborative tag suggestionCollaborative Tag Suggestions & Tag Spam Elimination Basic Algorithm

  13. Collaborative tag suggestion • Content-based Tag Suggestions • analysis and classification of the tagged content and context • Tag Normalization (improve tag uniformity, by computing the bi-grams of the tags in the currently chosen tag set) • stemming, edit distance, thesauri • Temporal Tags • time sensitive • Adjustments • dampening tag coverage score • adding coefficients to the penalizing and rewarding forces

  14. it also honors the popular demand ofusers three fairly orthogonal facets Examples

  15. Examples

  16. Conclusions • We discussed the basic criteria for a good tagging system and proposed a collaborative algorithm for suggesting tags that meet these criteria. • Our preliminary experience shows that a simple embodiment of such an algorithm is effective. • We have implemented a simplified tag suggestion scheme in My Web 2.0. • Our experience shows that this simple scheme is quite effective in suggesting appropriate tags that possess the properties proposed by us for a good tagging system.

  17. Future work • Develop metrics to quantitatively measure the quality of suggested tags, and study how tag suggestion can help to facilitate convergence of tag vocabulary • Introduce automatically generated content-based tags and also consider the time-sensitivity of tags. • Improve tag uniformity by normalizing semantically similar tags that are not similar in letters. The bi-gram method cannot achieve this. This would require incorporating certain linguistic analysis features. • Using voting and existing tags alone may prevent new high-quality tags from emerging.

  18. Comments • How a social network can be used to understand relationships between different taggers • How the co-occurrence of tags can be used to cover more facets of an object

More Related