The CoLiDeS Model

CoLiDeS stands for
Comprehension-based Linked model of Deliberate Search

A: Would you tell me which button to click?
B: This is it, CoLiDeS!

-- Demonstration of CoLiDeS --

Simulation of User Performing Book Browse Task at Amazon.Com

About the CoLiDeS Model

What is CoLiDeS

  • CoLiDeS (Comprehension-based Linked model of Deliberate Search) is a simulation model of navigation on the World Wide Web. CoLiDeS extends a series of earlier models developed by Kitajima & Polson (1995, 1997) for applications with GUIs, and the entire series of models is based on Kintsch's construction-integration theory of text comprehension (Kintsch, 1998). CWW (Cognitive Walkthrough for the Web) evaluates each web page by simulating user behavior using the CoLiDeS model.

Basics of CoLiDeS

    • A web page is made up of a large collection of objects competing for users' attention (e.g., action graphic, hypertext link, navigation bar item, or paragraph), which are meaningful units and/or targets for action. Users manage this complexity by a two-phased processes.
      • Attention Phase: Users segment the page into regions and focus on a region of the page.
      • Action Selection Phase: Users first comprehend each of the objects (e.g., hypertext link, graphic, radio button, etc.) that can be acted on in the focused-on region, including the consequences of acting on the object. Then they select one of the actions, usually clicking on one of the available hyperlinks.
    • The user's behaviors in the attention and the action selection phases are determined by the perceptions of similarity between the user's goals and the descriptions of alternative regions or actions.
    • These similarities are calculated by Latent Semantic Analysis (LSA).

The CoLiDeS papers:

About the Demo

The demo provides a step-by-step simulation of a user performing a browse-for-books task at Amazon.Com, by using four types of screen images:

  • New page: Actual page images that the website provides, which are input to the attention phase.
  • Segment page. Images showing how the user segments the web page into regions.
  • Focus on one region: Images that show which region the user focuses on for further processing, blurring the regions that the user is not focusing on and showing the focused-on region in clear font and images. Together the segmentation-into-regions process and focusing-on-one-region process constitute the attention phase.
  • Select action: After first comprehending all the available actions in the focused-on region, the user selects the one action that is most similar to the user's goal. Images show the region the user focuses on during the action selection phase, and an inserted arrow points to the action that the user selects.

A separate web page displays each of the 15 steps in the demo. An image from appears on the right side of the page, and a box in the upper left corner indicates the current processing status corresponding to the image. Below the box (to the left of the image) appears text explaining the current processing status and pointing out important features to notice in the image.

Click Here to Start Demo.

Created by Muneo Kitajima,