Table 7.
Mean average precision of nine term selection algorithms by choosing digit 20 as threshold to categories ranked corpus documents as relevant and non-relevant
| No. of terms | FR metrics |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Chi-square | Chi | DFC | KLD | RSV | CoDice | IG | LRF | PRF | |
| 5 | 0.2574 | 0.2614 | 0.2859 | 0.2925 | 0.2824 | 0.2727 | 0.2700 | 0.2835 | 0.2624 |
| 10 | 0.2668 | 0.2519 | 0.2892 | 0.2754 | 0.2657 | 0.2777 | 0.2666 | 0.2787 | 0.2651 |
| 15 | 0.2660 | 0.2531 | 0.2878 | 0.2600 | 0.2553 | 0.2714 | 0.2674 | 0.2660 | 0.2637 |
| 20 | 0.2616 | 0.2515 | 0.2890 | 0.2556 | 0.2495 | 0.2672 | 0.2677 | 0.2605 | 0.2663 |
| 25 | 0.2656 | 0.2514 | 0.2892 | 0.2522 | 0.2441 | 0.2657 | 0.2663 | 0.2587 | 0.2643 |
| 30 | 0.2745 | 0.2504 | 0.2910 | 0.2443 | 0.2381 | 0.2667 | 0.2635 | 0.2554 | 0.2591 |
| 35 | 0.2642 | 0.2496 | 0.2887 | 0.2406 | 0.2375 | 0.2620 | 0.2608 | 0.2530 | 0.2619 |
| 40 | 0.2613 | 0.2482 | 0.2912 | 0.2391 | 0.2381 | 0.2598 | 0.2580 | 0.2495 | 0.2598 |
| 45 | 0.2630 | 0.2452 | 0.2891 | 0.2358 | 0.2380 | 0.2601 | 0.2598 | 0.2476 | 0.2577 |
| 50 | 0.2630 | 0.2460 | 0.2903 | 0.2343 | 0.2338 | 0.2586 | 0.2569 | 0.2441 | 0.2565 |