I thought I’d share the work I carried out as part of my MSc dissertation in Intelligent Web Technologies.
The system has been implemented using Wordpress. A simplified version of this summarization system will be released as a Wordpress plugin in the following days. You can download the pdf version of my research work here: Incremental Personalised Summarisation with Novelty Detection
In recent years personalised summarisation has been a hot topic in Information Retrieval. An example of a personalisation technique is query-biased summarisation, where the importance of a sentence is estimated based on the amount of information about the query that it contains. An appealing implementation of this method is the one undertaken by many web search enginesthat provide snippets generated with sentences, or part of sentences, that contain query terms. However, the query string is not enough to represent the user’s interests and information need, and a more complex representation is needed. We present in this work a method to create a richer user model by analyzing the interaction of the user with a blogging system. This model is then exploited to create personalised summaries that contain information tailored to the user’s interests. The system also allows the user to extend initial summaries with broader summaries that contain novel information, and dynamically adapt to the model of the user’s information need. A system- centric and user-centric evaluation suggest that the proposed summarisation method performs better than query-biased summarisation in extracting relevant sentences from a document, thus showing that adaptive summarisation is an effective way to support the user in understanding the relevancy of a document to a particular topic.