MTAF: An Adaptive Design for Keyword-Based Content Dissemination on DHT Networks

Beyond offering the widely used keyword search function, many peer-to-peer systems nowadays support the subscription function. For example, Vuze allows users to create subscription filters based on the keyword search. Given the subscription, episodic or related content will be delivered to the users whenever new episodes are available. Unfortunately, these applications suffer from the downsides, for example, high network traffic in the nodes maintaining popular terms. In this paper, we propose the MTAF mechanism to overcome the issues. The key of MTAF is to carefully select a subset of terms without incurring false negatives and to forward the content item toward the home nodes of such selected terms for low content forwarding cost. Experimental results based on real datasets indicate that the proposed solutions are efficient compared to existing approaches. In particular, the similarity-based replication of filters is shown to mitigate the effect of hot spots that arise due to the fact that some document terms are substantially more popular than the others.