Indexing reduction

In order to determine the behaviour of document indexing reduction on CLIRS, we submitted to the contest, a set of five runs, which are described as follows.

First Run: $\rightarrow$ Full:

This run used ``Full documents'' as evaluation corpus, and conformed the baseline for our experiments.

Second Run: $\rightarrow$ TP10:

This run used an evaluation corpus composed by the reduction of every document, using the TP technique with a neighbourhood of 10% around TP.

Third Run: $\rightarrow$ TP20:

This run used an evaluation corpus composed by the reduction of every document, using the TP technique with a neighbourhood of 20% around TP.

Fourth Run: $\rightarrow$ TP40:

This run used an evaluation corpus composed by the reduction of every document, using the TP technique with a neighbourhood of 40% around TP.

Fifth Run: $\rightarrow$ TP60:

This run used an evaluation corpus composed by the reduction of every document, using the TP technique with a neighbourhood of 60% around TP.

Table 1 shows the size of every evaluation corpus used, as well as the percentage of reduction obtained for each one. As can be seen, the TP technique obtained a big percentage of reduction (between 75 and 89%), which also implies a reduction in time for the indexing process in a CLIRS.

Table 1: Evaluation corpora

Corpus	Size (Kb)	% of Reduction
Full	117,345	0%
TP10	12,616	89.25%
TP20	19,660	83.25%
TP40	20,477	82.55%
TP60	28,903	75.37%