SearchAggregation techniques
“The actual success of a meta-search engine directly depends on the aggregation technique underlying it.“1 . In choosing an aggregation technique for the meta search project there were two main factors (i) How relevant would the aggregation results be (ii) How difficult would it be to implement the chosen aggregation technique. The difficulty lies in that it is hard(if not impossible) to know whether a technique is difficult to implememt before actual implementation and the same can be said for relevancy, which may be subjective. Both of these factors are best realised in comparison with the results of other aggregation techniques. Markov Chains:A Markov chain, is a mathematical system that undergoes changes from one state to another, between a number of possible states. It is described as memoryless: the next state depends only on the current state and not on the preceding events. It is a “random process in which the future is independent of the past, given the present”2. This specific kind of "memorylessness" is called the Markov property. Markvo chains are very effective but are difficult to conceptualize and implement in a system compared to other models. Bayes Fuse : The Bayes fuse is a rigourous yet simple model for Information retrieval. It requires only simple training training, and ranks as opposed to relevance scores 3. Like the Borda fuse however, it tends to lose on in terms of performance compared to more sophisticated models. Kemeny-Young aggregation:The Kemeny Young method is based upon the Condorcet principal. (The condorcet principal basically states that if a majority of voters prefer A than B then the aggregate supports A.) The Kemeny–Young method uses preferential ballots on which voters rank choices according to their order of preference. A voter is allowed to rank more than one choice at the same preference level. Unranked choices are usually interpreted as least-preferred. Finding a Kenny Optimal Aggregation is...