Authors: Sebastian Schmidt, Ines Zelch, Janek Bevendorff, Benno Stein, Matthias Hagen, Martin Potthast

Published on: February 07, 2024

Impact Score: 8.3

Arxiv code: Arxiv:2402.04889

Summary

• What is new: The novel use of LLMs as a countermeasure to block ads integrated within conversational search engine responses.
• Why this is important: The integration of advertising within the answers provided by conversational search engines, using LLMs, poses a subtle and manipulative advertising issue.
• What the research proposes: Using fine-tuned sentence transformers and state-of-the-art LLMs to detect and block ads integrated within conversational search engine answers.
• Results: Sentence transformers achieved high precision and recall values above 0.9 in detecting integrated ads, while LLMs struggled with this task.

Technical Details

Technological frameworks used: Sentence Transformers, Large Language Models (LLMs)

Models used: Fine-tuned sentence transformers

Data used: Large dataset of ad-prone queries and generated answers with automatically integrated ads

Potential Impact

Conversational search engines, advertising industry, companies utilizing native advertising and product placement strategies

Want to implement this idea in a business?

We have generated a startup concept here: AdBlockAI.