Semantic Routing:

Semantic Routing is a method of routing which is more focused on the nature of the query to be routed than the network topology. Essentially semantic routing improves on traditional routing by prioritising nodes which have been previously good at providing information about the types of content referred to by the query.

In order to be able to search for information on a p2p network semantically the data needs to have a semantic description associated with it, one popular solution is the use of RDF meta-data[1] for this purpose. Tagging documents/data with RDF would provide a rich 'semantic web' which could be structured in a p2p fashion. A schema-based p2p network such as this would benefit greatly from semantic routing. Semantic routing differs fundamentally from other routing techniques because prospective nodes are selected because of another node’s confidence in their ability to respond correctly to a given query irrespective of their position within the network.

There have been several different projects started with a view to examining the power of semantic routing, for instance: Neurogrid[2], Nejdl,Wolpers,Siberski et al[2] and Tempich,Staab,Wranik[4]. Arguably the most interesting of these is the Remindin[4] system which incorporates a semantic routing algorithm developed with the intention of mimicking social networks. In addition to this many of the techniques developed for collaborative filtering[5] are equally applicable to the ranking of peers within a semantically routed network.

Each time a node answers a query its peers adjust their confidence in that node with regard to that type of query. The nature of this adjustment depends upon whether the node answered correctly i.e. if the search result was selected by the searcher. The information associated with ‘types’ depends greatly on the kind of semantic data being dealt with by the network and the strictness of the peer confidence ranking algorithm. How this data is stored and how the semantic routing algorithm uses this data is beyond the scope of this article, however it has been shown[4] that automated relaxation of queries can lead to more robust searches.

An important factor in order for the routing algorithm to be effective in the long term is persistence. Nodes must have a constant identifier within the network’s namespace if they are to retain their confidence ratings. There is an initial forming stage where none of the peers have ratings for any nodes, and nodes might be returned randomly (or indeed using traditional routing methods). It has been observed[5] that a certain amount of random responses to all requests (even after the forming stage) can avoid an effect called ‘overfitting’, which is when the confidence data associated with the nodes becomes too rigid and inflexible.

Semantic routing is a reasonably new idea and as such is in its infancy as a technique. A universal consensus on a specific algorithmic approach has not been reached and there is not yet a paradigm under which research into p2p semantic routing is conducted. The advent of semantic networking is making this an important topic, and one that is sure to grow and flourish in future.

Semantic Routing

[1] W3C RDF page

[2] Neurogrid

[3] Super-peer-based routing and clustering strategies for RDF-based peer-to-peer networks

[4] Remindin

[5] Resources on Collaborative Filtering