Merging Results by Predicted Retrieval Effectiveness

In this paper we propose several merging strategies to integrate the result lists of each intermediate run in distributed MLIR. The prediction of retrieval effectiveness was used to adjust the similarity scores of documents in the result lists. We introduced three factors affecting the retrieval effectiveness, i.e., the degree of translation ambiguity, the number of unknown words and the number of relevant documents in a collection for a given query. The results showed that the normalized-by-top-k merging with translation penalty and collection weight outperformed the other merging strategies except for the rawscore merging.