Skip to main content
. Author manuscript; available in PMC: 2014 Jun 26.
Published in final edited form as: Open Inf Syst J. 2013 Nov 21;6:1–12. doi: 10.2174/1874133920131121001

Table 1.

Repeated training and addition of new pairs satisfying lexSim ≥ 0.5 based on the new weights obtained until convergence of the process. IDF produces the first total set of 141.2 Million pairs, then each wi is trained on the total pairs listed in its row and is used to produce the additional pairs added in the next row.

Weights Pairs added with lexSim ≥ 0.5 Total Pairs Pairs with conSim >0 Pairs with conSim ≤0
IDF, w0 141.2 M 41.9 M 99.2 M
w1 81.9 M 223.1 M 63.8 M 159.2 M
w2 181.7 M 404.8 M 106.5 M 298.3 M
w3 27.1 M 431.8 M 113.9 M 318.0 M
w4 8.5 M 440.4 M 116.4 M 324.0 M
0.6 M 441.0 M 116.8 M 324.2 M