Tracking the Navigation Behavior of Web Communities
- Presented by: Holger Wagner
- Supervisors: Prof. Dr. François Bry, Dipl.-Inform. Michael Kraus
- Date: 18-06-2001
About this Presentation
- Introduction to upcoming Project Thesis
- Presentation duration: approximately 20 minutes
- Intentions:
- Provide a broad overview of the main ideas
- Talk about the scope of project thesis
Main Ideas
- The Main Ideas behind the Project
- Give the abstract structure(s) of the Web a concrete face
- Know which users are in which "area"
- Add semantics to "areas"
- Let users know and easily reproduce their session history
- Let authors know how their content is used by which groups
- Provide more effective ways of searching in a limited Web
- Support collaborative Web research for (distributed) groups
- Integrate services to enhance the "Web experience"
- Support “nearby user chat” (distance = # of links in between)
- Make user comments / annotations available to others
- When the work is done: have some fun!!!
Example Application 1
- Tracking the Navigation Behavior...
- ... of research teams (within a particular project)
- Improve communication within the team
- Provide an easy way for team members to find out which documents have already been viewed by which participant including himself
- Let team members rate relevance of documents to project
- Find out which searching methods were most successful
- Create information useful to others
- Teams or individuals with similar interests may use resulting artifacts as starting point
Example Application 2
- Tracking the Navigation Behavior...
- ... of students within a particular academic content
- E.g.: Pages of the department of computer science
- E.g.: Pages of particular lectures, projects, seminars
- Find out how exactly the content is used by which groups
- Improve static structure of the content with this information
- Automatically generate dynamic alternative navigation structures based on user behavior
- Bring together people browsing same content so that they can collaborate (e.g. chat with viewers of same page)
Contents
- Terminology
- Web Communities
- Navigation Behavior
- Aspects of Tracking
- Semantic Modeling
- Grouping and Searching Documents
- Semantic User Profiles
- Visualizing the Results
- Test Bed for a Prototype
What is a Web Community?
- Groups of people visiting or looking for the same web pages or semantic groups of web pages
- Communities...
- ... may have different motivations for Web use
- ... may use the Web in different ways
- ... may be defined in different ways, resulting in...
- ... sharp, simple and well-defined bounds
- ... fuzzy bounds
- ... may exist in different temporal manners...
- ... from a moment to the life-time of its members
Examples of Web Communities
- Members of a department / teaching unit
- Students with a particular major
- People working on a particular (distributed) project
- People with similar interests or demographic backgrounds
- People currently looking for / browsing same information
- People with similar paths on the web
What is Navigation Behavior? (1)
- Navigation Behavior – Questions to be asked:
- What is the context of a Web session?
- Is the user involved in a specific action? E.g.:
- Research for a particular project
- Shopping
- Browsing news
- Killing time
- Are other documents viewed at the same time? Which?
- Which "areas" are viewed?
- Which semantic groups of documents?
- Which documents?
- Which sections within documents?
What is Navigation Behavior? (2)
- Navigation Behavior – Questions to be asked:
- Which hyperlinks are followed?
- How are hyperlinks followed?
- If user can choose: which modes?
- Replace current document with hyperlink target
- Show hyperlink target in new window
- Add content of hyperlink target where the hyperlink source was
- Only forwards or back and forth?
- How much time is spent in a particular area?
- Is the same area visited multiple times / regularly?
Aspects of Tracking
- How can the above questions be answered?
- Collect information on people's Web activity!
- Possible "Entry Points"
- Analysis of existing log files (server, client, whatever)
- On the information server
- Somewhere between server and client (e.g. at ISP)
- At the client computer
- Between browser software and Internet (local proxy)
- As a plugin / add-on for an existing browser software
- With an own implementation of a new browser software
Semantics for Documents
- Adding Semantics to Documents
- Rationale behind this:
- Provide intelligent automatic grouping of documents
- Make better searching techniques possible
- Possible sources of annotations for documents or sections thereof:
- Visitors / Users / "The Community"
- Authors / Content providers
- Related Work: Semantic Web
Semantic User Profiles
- Users may "classify themselves" with annotations
- Profiles can be matched via some sort of distance function
- Sections of profiles may be used to match specific interest groups
Annotations vs. Comments
- Comments are human-only understandable
- Made by humans, made for humans
- Annotations should be machine-processable
- Made by humans or machines, made mostly for machines
- Machines should be able to process annotations in some meaningful way, e.g. for...
- ... grouping of entities
- ... generating new knowledge
- Annotations should be easy to visualize in a human-understandable way
Semantic Modeling
- What could be used?
- Keywords, possibly structured in some way
- Ontologies, for example defined with:
- Resource Description Framework (RDF)
- Simple HTML Ontology Extensions (SHOE)
- Related Work: W3C Semantic Web Activity
- Objective: enhancing the web with semantics
Visualizing the Results
- Static vs. Interactive Visualization
- Visualizing a part of or the complete "visited Web"
- Visualizing the trail(s) of a particular user / community
- Visualizing the semantic model (demo of an example)
- Visualizing annotations
- Visualizing the "Community"
- Visualizing "nearby users" / users on same page
- Show number of other users currently "in the neighborhood"
- Enhancing the documents
- Highlighting "most often followed links" on a page
- Highlighting keywords that have to do with annotations
Test Bed: Community / Content
- Users of our computer pool
- A well-known community
- Students
- Staff
- Participants of projects involving research
- Content is not restricted
- Collaborative Research capabilities can be tested
- Users of the content offered by the department
- People interested in the Department of Computer Science of the University of Munich (DCSUM)
- Well-known and understood content
Project Thesis