# Predictive Web Prefetching: A Combined Approach Using Clustering Algorithms and WEKA in High-Traffic Settings

**Authors:** Adeyimi Abel Ajibesin (American University of Nigeria, Yola, Nigeria) — ORCID 0000-0001-6518-0231; Narasimha Rao Vajjhala (University of New York Tirana, Tirana, Albania) — ORCID 0000-0002-8260-2392; Ernest Joel (American University of Nigeria, Yola, Nigeria) — ORCID 0000-0001-6177-0090; Sandip Rakshit (RMIT Vietnam, Hanoi, Vietnam) — ORCID 0000-0003-2495-6771
**Type:** Conference paper
**Source:** Artificial Intelligence in Internet of Things (IoT): Key Digital Trends: Proceedings of 8th International Conference on Internet of Things and Connected Technologies (ICIoTCT 2023), Lecture Notes in Networks and Systems, vol. 1072, pp. 221–231, Springer Nature Singapore
**Published:** 2024-10-17
**DOI:** https://doi.org/10.1007/978-981-97-5786-2_17
**Canonical page:** https://www.narasimharao.net/research/predictive-web-prefetching-clustering-weka-high-traffic/
**Indexing:** Scopus
**Keywords:** Predictive web prefetching; Network congestion; Clustering algorithms; WEKA toolkit; High-traffic environments; Online navigation graph; Server idle times; Web caching synergy; Domain client groups
**Methodology:** Design and evaluation of a web prefetching model: preprocessing of server log files, construction of an online navigation graph, clustering with the WEKA machine learning toolkit, and comparison against conventional prefetching techniques on domain client-group logs.

## Research summary

- **The Problem:** Conventional web prefetching assumes servers have idle time, an assumption that breaks down in high-traffic, congested network environments.
- **The Methodology:** Design and evaluation of a web prefetching model: preprocessing of server log files, construction of an online navigation graph, clustering with the WEKA machine learning toolkit, and comparison against conventional prefetching techniques on domain client-group logs.
- **The Core Finding:** A prefetching approach that builds an online navigation graph from server logs and clusters navigation behaviour with WEKA outperformed conventional prefetching techniques on domain client-group logs, particularly once server idle time allowed prefetching across all user predictions.
- **The Citation:** Ajibesin, A. A., Vajjhala, N. R., Joel, E., & Rakshit, S. (2025). Predictive Web Prefetching: A Combined Approach Using Clustering Algorithms and WEKA in High-Traffic Settings. In Frank Lin, David Pastor, Nishtha Kesswani, Ashok Patel, Sushanta Bordoloi, Chaitali Koley (Eds.), Artificial Intelligence in Internet of Things (IoT): Key Digital Trends: Proceedings of 8th International Conference on Internet of Things and Connected Technologies (ICIoTCT 2023) (pp. 221–231). Springer Nature Singapore. https://doi.org/10.1007/978-981-97-5786-2_17

## Summary in detail

**Question.** How can web prefetching be improved for high-traffic, congested network environments in which servers have little idle time?

**Finding.** The authors propose a prefetching approach that builds an online navigation graph from preprocessed web server log data and uses clustering (with the WEKA toolkit) to predict the next related web items a user will request. Evaluated on log files from specific domain client groups, the clustering method outperformed conventional prefetching techniques, particularly after a short-lived acceleration phase once server idle time was sufficient to prefetch across all user predictions.

**Why it matters.** Network congestion reduces the usefulness of information because pages load slowly. A prefetching method that works when server idle time is scarce, and that combines with web caching, is directly applicable to busy web services and IoT-connected environments.

## Key findings

1. A prefetching approach tailored to high-traffic environments with minimal server idle time is introduced; it builds an online navigation graph from preprocessed log data to capture user navigation patterns.
2. Clustering algorithms, implemented through the WEKA toolkit, are used to group navigation behaviour and predict subsequent requests.
3. On log files from specific domain client groups, the proposed method surpassed conventional prefetching techniques, especially after a short-lived acceleration phase once server idle time was sufficient for prefetching across all user predictions.
4. When combined with web caching, the method can predict the related web items likely to follow a particular web object request.

## Study at a glance

| Item | Detail |
|---|---|
| Problem | Network congestion degrades Internet performance; prefetching in high-traffic settings with minimal server idle time |
| Design | Model development and empirical evaluation on web server log files |
| Data | Log files from specific domain client groups (preprocessed) |
| Methods | Online navigation graph of user navigation patterns; clustering algorithms via WEKA; combination with web caching |
| Main result | The proposed clustering method surpasses conventional prefetching techniques after a short acceleration phase when idle time allows prefetching for all predictions |
| Implication | Prefetching combined with caching can predict subsequent related web items after a given request |

## Abstract

Network congestion poses challenges for internet users, diminishing the utility of accessed information due to reduced speeds. To mitigate this, various strategies such as web mining, caching, and server prefetching have been proposed to enhance internet performance. We introduced a novel prefetching approach tailored for high-traffic environments with minimal server idle times. This method constructs an online navigation graph from preprocessed log data, offering insights into user navigation patterns across the web. Our primary objective was to refine web prefetching techniques for congested digital landscapes. We evaluated the model using log files from specific domain client groups. Our proposed web clustering method surpasses conventional prefetching techniques, especially after a short-lived acceleration when server idle times are sufficient for prefetching across all user predictions. This method, when synergized with web caching, can predict subsequent related web items following a particular web object request.

## When this research may be relevant

This paper may be relevant to researchers working on web prefetching and caching, web usage mining and navigation-pattern prediction, clustering-based prediction with WEKA, and performance of web services under congestion or in IoT settings.

## Limitations

- The evaluation used log files from specific domain client groups; the abstract does not report results for other traffic profiles.
- The advantage is described as emerging after a short-lived acceleration phase, i.e. it depends on sufficient server idle time.

## How to cite

Ajibesin, A. A., Vajjhala, N. R., Joel, E., & Rakshit, S. (2025). Predictive Web Prefetching: A Combined Approach Using Clustering Algorithms and WEKA in High-Traffic Settings. In Frank Lin, David Pastor, Nishtha Kesswani, Ashok Patel, Sushanta Bordoloi, Chaitali Koley (Eds.), Artificial Intelligence in Internet of Things (IoT): Key Digital Trends: Proceedings of 8th International Conference on Internet of Things and Connected Technologies (ICIoTCT 2023) (pp. 221–231). Springer Nature Singapore. https://doi.org/10.1007/978-981-97-5786-2_17

```bibtex
@inproceedings{ajibesin2025predictive,
  title = {Predictive Web Prefetching: A Combined Approach Using Clustering Algorithms and WEKA in High-Traffic Settings},
  author = {Ajibesin, Adeyimi Abel and Vajjhala, Narasimha Rao and Joel, Ernest and Rakshit, Sandip},
  booktitle = {Artificial Intelligence in Internet of Things (IoT): Key Digital Trends: Proceedings of 8th International Conference on Internet of Things and Connected Technologies (ICIoTCT 2023)},
  series = {Lecture Notes in Networks and Systems},
  editor = {Frank Lin and David Pastor and Nishtha Kesswani and Ashok Patel and Sushanta Bordoloi and Chaitali Koley},
  pages = {221--231},
  year = {2025},
  publisher = {Springer Nature Singapore},
  doi = {10.1007/978-981-97-5786-2_17},
  url = {https://doi.org/10.1007/978-981-97-5786-2_17}
}
```