Information Technology Reference
In-Depth Information
Table 10.2
(Continued)
ID
Feature description
40
LMIR.JM of the whole document
41
Sitemap based term propagation
42
Sitemap based score propagation
43
Hyperlink based score propagation: weighted in-link
44
Hyperlink based score propagation: weighted out-link
45
Hyperlink based score propagation: uniform out-link
46
Hyperlink based feature propagation: weighted in-link
47
Hyperlink based feature propagation: weighted out-link
48
Hyperlink based feature propagation: uniform out-link
49
HITS authority
50
HITS hub
51
PageRank
52
HostRank
53
Topical PageRank
54
Topical HITS authority
55
Topical HITS hub
56
Inlink number
57
Outlink number
58
Number of slash in URL
59
Length of URL
60
Number of child page
61
BM25 of extracted title
62
LMIR.ABS of extracted title
63
LMIR.DIR of extracted title
64
LMIR.JM of extracted title
•
Raw information of the documents associated with each query, such as the term
frequency and the document length.
•
Relational information, such as the hyperlink graph, the sitemap information, and
the similarity relationship matrix of the corpus.
With the meta information, one can reproduce existing features, tune their pa-
rameters, investigate new features, and perform some advanced research such as
relational ranking [
14
,
15
].
10.6 Learning Tasks
The major learning task supported by LETOR is supervised ranking. That is, given
a training set that is fully labeled, a learning-to-rank algorithm is employed to learn
