Comparison with Oracular Source selec tion We compared TSR with oracular source selection, DSR described above in Section 7.3.1. We compared TSR(0.9) with DSR(0.9) (i.e. linear combination 0.1 × DSR + 0.9 × XXXX). As shown in Figures 5(a) and 5(b), TSR(0.9) is able to match DSR(0.9) per- formance for the test queries. The aggregate results across the topics is shown in Figure 5(a) and topic- wise result is shown in Figure 5(b). Result shows that the TSR precisions are quite comparable with that of DSR. This implies that TSR is highly effective in cat- egorizing sources and queries, almost matching with oracular DSR. A note on the DSR’s performance for camera-topic. After investigating our deep-web envi- ronment for camera-topic, we found that the source- rank for camera-topic was dominated by sources which answered less than 25% of sampling queries. This could be attributed to the fact that our source selection technique led to selection of relatively more number of cross-topic sources than pure sources for camera topic. As a result, selecting top-ranked camera-topic sources infact led to a drop in performance.
