Research Article
Ranking of Search Requests in the Digital Information Retrieval System Based on Dynamic Neural Networks
Table 3
Results of experimental comparison of the efficiency of problem solving with the number of ranked links from 20 to 200.
| Ranking algorithm | CPU run time (s) | 20 links | 80 links | 200 links |
| Vector neural network ranking | 5,1667е − 004 | 2,1667е − 002 (10 model runs) | 0,1211 (5 model runs) | Scalar neural network ranking | 6,379е − 004 | 9,8623ее − 002 (10 model runs) | 0,5274 (5 model runs) | Sorting the relevance of found links by the “bubble” method | 1,1437е − 003 | 0.5274 | 8.4578 | Sorting the relevance by selection of the largest element | 1,3496е − 003 | 0.8305 | 9.6424 | Ranking by sorting the array of values by insert | 2,0054е − 003 | 0.9846 | 11.2836 | Exhaustive search | 10.2159 | — | — |
|
|