Semantic Web Site Usage Analysis: The ORGAN system

1
Semantic Web Site Usage Analysis The ORGAN
system

• John Garofalakis Theodoula Giannakoudi
  Evangelos Sakkopoulos

RA Computer Technology Institute Internet and
Multimedia Technologies RU 5 and WestGate N.
Kazantzaki str. 26504, Greece
University of Patras Computer Engineering
Informatics Dept 26500 Patras, Greece
2
ORGAN overview

• ORGAN Ontology-oRiented usaGe Analysis system
• Web site usage analysis system by using semantic
  knowledge
• Traditional usage Web site domain
• analysis
  knowledge
• combined semantically enhanced queries
  about the web site usage
• Available integrated and standalone system

3
System architecture
4
Keywords Extraction

• Content parsing,HTML
• tags removal and
• indexing
• Removal of stop-words
• Calculation of words
• incidence
• Extraction of top most
• frequent keywords

• Spotting of the area
• around the specific link
• in the HTML code,
• which is anchored
• 100 bytes before the link
• and 100 bytes
• after the link.

5
ORGAN Translation

• Translation of non-English written sites web page
  keywords
• to facilitate lexical processing
  (through WordNet, for
  words of the English vocabulary only)
• Automatic keywords translation through a local
  web service

6
ORGAN Metadata Assigner

• Calculation of semantic similarity
• WordNet electronic lexical database for words
  of the English vocabulary only following an
  hierarchical concepts organization
• Wu Palmer measure calculates relatedness by
  considering the depths of the two synsets (one or
  more set of synonyms) in the WordNet taxonomies
• score 2depth(lcs) / (depth(s1) depth(s2))
• where lcs root taxonomy, depth 1
• s1,s2 synsets.
• 0ltscore1

7
ORGAN LogPro

• Log files preprocessing
• Data cleaning from records of requests for
  images, requests for non-informational files
• Data cleaning from records of requests that were
  not successfully responded or were submitted by
  search spiders robots
• Sessions identification based on the user IP,
  the user agent and the time interval between
  subsequent requests.
• Path completion - filling the page references
  that are missed due to local browsing caching
  mechanisms

8
ORGAN Analyzer (1)

• Query builder that mines logs with respect to
  site semantics
• Ontology querying functionality Protégé API
• Analysis of knowledge derived from three
  resources
• Web server raw log files
• Metadata information of the site
• OWL ontology
• Queries type

9
ORGAN Analyzer (2)
10
Conclusion Future Work

• ORGAN integrated tool with a service-oriented
  architecture providing web site usage semantic
  log analysis.
• Future steps
• more sophisticated keywords extraction
• enhancement of log file processing
• Support of web site reorganization tasks

11
Thank you for your attention!