A method for predicting click rate of a Wikipedia hyperlink based on webpage features

By extracting text, image, and visual features and combining them with a ranking learning algorithm, the problem of visual factors not being considered in existing technologies is solved, and high-accuracy hyperlink click-through rate prediction is achieved under different devices and resolutions.

CN116226558BActive Publication Date: 2026-06-12SHANGHAI JIAOTONG UNIV
2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHANGHAI JIAOTONG UNIV
Filing Date
2022-11-18
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

Existing technologies fail to effectively consider visual factors when predicting click-through rates for Wikipedia hyperlinks, resulting in low accuracy. Furthermore, existing visual feature schemes lack universality and cannot adapt to browsing scenarios with different devices and resolutions.

⚗Method used

We extract text features, graph-based features, and visual features, and combine them with a ranking learning algorithm to predict click-through rates. In particular, we obtain text similarity and visual features through a pre-trained language model and a community detection algorithm, and use a decision tree-based list ranking algorithm to predict hyperlink click-through rates.

🎯Benefits of technology

It improves the accuracy of hyperlink click-through rate prediction, is applicable to browsing scenarios with different devices and resolutions, eliminates the impact of the number of hyperlinks on click-through rate, and enhances the universality and accuracy of prediction results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
Patent Text Reader

Abstract

The application relates to a Wikipedia hyperlink click rate prediction method based on webpage features, which comprises the following steps: extracting text features, graph-based features and visual features from hyperlinks in a Wikipedia webpage; and outputting a hyperlink click rate prediction result according to the extracted text features, graph-based features and visual features and combining a ranking learning algorithm. Compared with the prior art, the application fully considers the influence of visual factors on the hyperlink click rate, and proposes visual features suitable for all scenes, can effectively improve the accuracy of the prediction result by extracting text features, graph-based features and visual features from the hyperlinks and combining the ranking learning algorithm for the click rate prediction.
Need to check novelty before this filing date? Find Prior Art