Vogt, C. When Does It Make Sense to Linearly Combine Relevance Scores? Poster in Proceedings of ACM SIGIR (Philadelphia, PA, 1997).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....We used trec eval to generate the performance results. Our multi channel merging results were produced using the combSUM[5,11] approach, that is, we summed the similarity values for images across the channels in which the image was included in the response set. The conditions set out by Vogt[14] for linearly combining relevance scores apply here: our channels will be seen to have reasonable performance and they do not rank relevant documents similarly. We made no attempt to optimize the merging algorithm although intuition suggests that a weighted sum is almost certainly appropriate. ....

Vogt, C. When Does It Make Sense to Linearly Combine Relevance Scores? Poster in Proceedings of ACM SIGIR (Philadelphia, PA, 1997).

....the source specific system was weighted more heavily. Considering this, the similar performance of the source specific system and the combined system is not surprising, but raises the question of why one system is always weighted more heavily than the other. Work which we have done concurrently ([Vogt, 1997]) may shed some light on this it appears that the problem is which systems we chose to combine. Our work indicates that the best time to linearly combine systems is when they a) both have performance of similar magnitudes and b) rank relevant documents differently. For our training data, nearly ....

....linear model, we are investigating neural network models which are capable of implementing a broader range of combination functions and taking query and document representations into account. 1 As measured by GPAr , the Guttman s Point Alienation calculated using only relevant documents see [Vogt, 1997] Perhaps the most interesting conclusion from this work is that our training technique this year, which included per query training, was robust to a poor selection of component systems. Given a pair of experts which were unlikely to be combinable, the training process was able to identify the ....

Vogt, C. C. (1997). When does it make sense to linearly combine relevance scores? In [Belkin et al., 1997]. Poster Session. UCSD CSE Tech Report CS97-556.

No context found.

Vogt, C. When Does It Make Sense to Linearly Combine Relevance Scores? Poster in Proceedings of ACM SIGIR (Philadelphia, PA, 1997).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC