Presentation is loading. Please wait.

Presentation is loading. Please wait.

A Method for Analyzing User Action Logs Center for E-Business Technology Seoul National University Seoul, Korea Jaeseok Myung Intelligent Database Systems.

Similar presentations


Presentation on theme: "A Method for Analyzing User Action Logs Center for E-Business Technology Seoul National University Seoul, Korea Jaeseok Myung Intelligent Database Systems."— Presentation transcript:

1 A Method for Analyzing User Action Logs Center for E-Business Technology Seoul National University Seoul, Korea Jaeseok Myung Intelligent Database Systems Lab School of Computer Science & Engineering Seoul National University, Seoul, Korea Papers - Interception of User's Interests on the Web - User Characteristics Acquisition from Logs with Semantics - Estimation of User Characteristics using Rule-based Analysis of User Logs - Personalized Presentation in Web-Based Information Systems

2 Copyright  2008 by CEBT About Project NAZOU  Tools for acquisition, organization and maintenance of knowledge  NAZOU is proposed by FIIT in Slovakia University to realize the slogan  The result of this research are verified by pilot application about job offers  http://nazou.fiit.stuba.sk/home/index.php http://nazou.fiit.stuba.sk/home/index.php  Paper Info. Barla, M.: Interception of User's Interests on the Web. In V. Wade, H. Ashman, and B. Smyth, editors, 4th Int. Conf. on Adaptive Hypermedia and Adaptive Web-Based Systems, AH'06, pages 435-439, Du blin, Ireland, 2006, Springer, LNCS 4018Interception of User's Interests on the Web. Barla, M., Andrejko, A., Bieliková, M., Tvarožek, M.: User Characteristics Acquisition from Logs with S emanticsIn ISIM ´07 Information Systems and Formal Models: 10th International Conference on Inform ation System Implementation and Modeling. 2nd International Workshop on Formal Models WFM ´07 2007, Hradec nad Moravicí, Czech Republic, pp.103-110, 2007User Characteristics Acquisition from Logs with S emantics Barla, M., Bieliková, M.: Estimation of User Characteristics using Rule-based Analysis of User Logs. In : Data Mining for User Modeling Proceedings of Workshop held at the International Conference on Use r Modeling UM2007, Corfu, Greece, pp.5-14, 2007Estimation of User Characteristics using Rule-based Analysis of User Logs. Tvarožek, M., Barla, M., Bieliková, M. (2007). Personalized Presentation in Web-Based Information Sy stems. In J. Van Leeuwen, G. F. Italiano, W. van der Hoek, H. Sack, C. Meinel, F. Plášil (Ed.), SOFSE M 2007: Proceedings of the 33rd Conference on Current Trends in Theory and Practice of Computer S cience. LNCS 4362, pp. 796-807. Harrachov, Czech Republic: Springer-Verlag, Berlin Heidelberg.Personalized Presentation in Web-Based Information Sy stems. IDS Lab Seminar (User Log Analysis) - 2

3 Copyright  2008 by CEBT Architecture of NAZOU  Goal : Personalized Knowledge Presentation IDS Lab Seminar (User Log Analysis) - 3

4 Copyright  2008 by CEBT Components of NAZOU IDS Lab Seminar (User Log Analysis) - 4

5 Copyright  2008 by CEBT Logging Issue(1)  Client-side Logging Captures events that occur in the web browser Captures individual actions which might be missed by server-side – Load, Unload, Click, MouseOver, MouseOut, … – Cached Page Loading Time Privacy problem might be happened if we use desktop apps Click – Technologies used : JavaScript, Ajax, SOAP – Input : user input in form of performed action – Output : Log of user actions sent to server – http://nazou.fiit.stuba.sk/home/?page=click http://nazou.fiit.stuba.sk/home/?page=click IDS Lab Seminar (User Log Analysis) - 5

6 Copyright  2008 by CEBT Logging Issue(2)  Server-side Logging Standard web server logs are not suitable for the estimation of individual user characteristics, since they require complicated preprocessing – 1Cust216.tnt1.santa-monica.ca.da.uu.net - -[08/May/1999:12:13:03 -0700] GET /gen/meeting/ssi/next/HTTP/1.0 200 9887 http://www.slac.stanford.edu/ Mozilla/3.01-C-MACOS8 (Macintosh; I; PPC) GET /gen/meeting/ssi/next/ - HTTP/ 1.0IPhttp://www.slac.stanford.edu/ We need to preserve the semantics of user action as possible as we can SemanticLog – Collaborating with Click (use pre-defined event ontology) – Technologies used : Java, XML, MySQL, Hibernate, Web Services – Input : Event ontology, Semantic events corresponding to user actions – Output : Integrated log of user events in a relational database corresponding to the event ontology schema – http://nazou.fiit.stuba.sk/home/?page=semantic-log http://nazou.fiit.stuba.sk/home/?page=semantic-log IDS Lab Seminar (User Log Analysis) - 6

7 Copyright  2008 by CEBT User Modeling  Ontology-based User Model used in NAZOU Project http://nazou.fiit.stuba.sk/home/files/nazou_um.pdf Stored as triples in RDF repository GUMO(The General User Model Ontology) IDS Lab Seminar (User Log Analysis) - 7

8 Copyright  2008 by CEBT Rule-based Estimation of User Char.  Knowledge on user characteristics acquisition is represented by rules IDS Lab Seminar (User Log Analysis) - 8

9 Copyright  2008 by CEBT Rule Definition  Each rule consists of Pattern Consequence  Pattern Defined as a sequence of event types and other sub-sequences  Consequence Determines what and how should be changed in the user model when the instance of a pattern is detected  Rules are stored in a file using XML format IDS Lab Seminar (User Log Analysis) - 9

10 Copyright  2008 by CEBT Pattern Example  http://fiit.sk#PropertyUri http://fiit.sk#restrictionURI http://fiit.sk#SelectRestrictionhttp://fiit.sk#SameAsPrevioushttp://fiit.sk#DifferentThanPrevious IDS Lab Seminar (User Log Analysis) - 10

11 Copyright  2008 by CEBT Pattern Detection  How can we find instances of appropriate patterns IDS Lab Seminar (User Log Analysis) - 11 DetectPattern(Event): find rule candidates for Event; for each rule in rule candidates find applicable rule instances(rule, event); for each rule instance in applicable rule instances apply event on rule instance; findRuleCandidates(Event): for each rule in known rules if type of event matches the first event of pattern part of the rule add rule to candidate rules; else if exists such rule instance of rule belonging to the current user that type of expected event match type of upcoming event add rule to candidate rules; return candidateRules; findApplicableRuleInstances(Rule, Event): for each ruleInstance of rule belonging to current user checkContextOfCurrentSequence(Event); checkContinuity(Event); checkContextOfEvent(Event); if all checks passed add ruleInstance to applicableRuleInstances; return applicableRuleInstances;

12 Copyright  2008 by CEBT User Model Update  Update of a user model is driven by changes specified in the consequence part of rule IDS Lab Seminar (User Log Analysis) - 12 UMupdate(): retrieveInstanceOfUserCharacteristic(); //which is being changed for each property in processed properties; update value according to given strategy; update timestamp; update count-of-updates; retrieveInstanceOfUserCharacteristic(): check value of all referencing properties; check value of all used properties; if rule does not allow for change of ``foreign characteristic check value of source of characteristic; if no instance fulfills these criteria create a new instance; set all referencing and used properties; set source; return found or created instance;

13 Copyright  2008 by CEBT Updated User Model in NAZOU IDS Lab Seminar (User Log Analysis) - 13

14 Copyright  2008 by CEBT Conclusion  Click Realize client-side logging Use Javascript technology (Good for Security Problem)  SemanticLog Implements server-side logging by means of a web service(SOAP) Aggregate one common log per user session Store logs in a relational database  LogAnalyzer Estimate user characteristics from the acquired user logs Evaluates incoming events and updates user model User characteristics can be used for personalized services IDS Lab Seminar (User Log Analysis) - 14

15 Copyright  2008 by CEBT Evaluation & Discussion Points  Pros Interesting Ideas & (Well-bounded & Well-positioned) Components Ontologies & Tools Implementation  Cons No experiments Poor examples Poor explanation  Discussion Points What kind of characteristics can we get from logs? How can we model those user characteristics? How can we estimate those characteristics? What are possible services from the user model? IDS Lab Seminar (User Log Analysis) - 15


Download ppt "A Method for Analyzing User Action Logs Center for E-Business Technology Seoul National University Seoul, Korea Jaeseok Myung Intelligent Database Systems."

Similar presentations


Ads by Google