=Paper=
{{Paper
|id=Vol-1171/CLEF2005wn-adhoc-LlopisEt2005
|storemode=property
|title=Combining Passages in the Monolingual Task with the IR-n System
|pdfUrl=https://ceur-ws.org/Vol-1171/CLEF2005wn-adhoc-LlopisEt2005.pdf
|volume=Vol-1171
|dblpUrl=https://dblp.org/rec/conf/clef/LlopisN05a
}}
==Combining Passages in the Monolingual Task with the IR-n System==
https://ceur-ws.org/Vol-1171/CLEF2005wn-adhoc-LlopisEt2005.pdf
Combining passages in monolingual task with
IR-n system
Fernando Llopis and Elisa Noguera
Grupo de investigación en Procesamiento del Lenguaje Natural y Sistemas de Información
Departamento de Lenguajes y Sistemas Informáticos
University of Alicante, Spain
llopis,elisa@dlsi.ua.es
Abstract
This paper describes our participation in monolingual tasks at CLEF-2005. In
this research we have worked in the following languages: English, French, Portuguese,
Bulgarian and Hungarian. Our task has been focused on using combined different
size passages to improve the Information Retrieval process. Once we have studied the
experiments which have been carried out and the official results at CLEF, we have
realized that this combining model gets better the achieved scores considerably.
Categories and Subject Descriptors
H.3 [Information Storage and Retrieval]: H.3.3 Information Search and Retrieval
General Terms
Experimentation, Measurement, Performance
Keywords
Information Retrieval
1 Introduction
Information Retrieval systems based on passages (PR) [2] determine the relevance of a document
regarding to a question. This relevance is obtained from the similarity of different fragments in
this document regarding to the same question. This models not only let to improve the location of
relevant documents, but also let us to find the most relevant part of the document accurately. This
last advantage allows us that these systems which are used in other tasks as Question Answering
(QA).
PR systems are classified according to how the passages are determined in each document.
IR-n system is a PR system which defines the passages based on a fixed number of sentences. This
provides the passages with some syntactical content. Last years our researches with IR-n system
are based on detecting the suitable size for each collection ( to experiment with test colletion ),
but determining the similarity of a document based on the passage with more similarity. This
year the score which is given to each document is based on the similarity of several size passages.
This paper is organized as follows: next section describes IR-n system and its new changes.
Following, we describe the task developed at CLEF 2005 by our system and the training. And
finally, we present the archieved results and the conclusions.
2 IR-n system
IR-n system [3] was developed in 2001. It was written in C++ entirely, although it has been used
external tools (stemmers) in few occasions. Last year [4] the system was designed and developed
again with the aim of getting more information (size of passages in words) and to improve the
process speed.
The system works in Linux system without excesive requeriments, nevertheless, in view of the
fact that the search process is carried out with structures load in memory, it is convenient that
the computer has the enough memory.
In this section is presented the main charateristics of IR-n system and it is detailed the combined
passages method used in this edition.
2.1 Resources: stemmers and stopword lists
This was the first year that we have worked with Bulgarian and Hungarian languages. It has
been used the stemmers and stopwords lists available on the web http://www.unine.ch/info/clef.
We can highlight that Hungarian and Bulgarian collections are encoded in UTF-8. In addition
Bulgarian stemmer is developed in perl to support UTF-8. The rest of the stemmers are developed
in C.
2.2 Similarity measures
IR-n system is ready for using several similarity measures: cosine [5], pivoted cosine [7] and okapi
[6]. To the last one the values of parameters (k1,b,avg) could be update in a easy way, in order to
get the best results.
On the whole, experiments carried out by okapy measure show us that we could obtain the
best results. Furthermore, we have contrasted normalization concept with size passage again.
Previous versions of IR-n system does not use size passage in the similarity measures because
all the passages had the same size. Last edition we could check that results improved if it was
considered size passages.
2.3 Query expansion
Most of IR systems use query expansion techniques [1] based on adding the most frequent terms
contain in the more relevant documents to the original query. Architecture IR-n allows us to use
query expansion based on more relevant passages or documents. In fact, last edition we got better
results using the more relevant passages.
2.4 Combined passages
The present year, technique called ’combined passages’ has been developed. The model consists
of applying similar techniques for merging relevant document lists in multilingual task but using
relevant passage lists of different size.
This model consists of using different size passages in order to get relevant document lists. The
list which have been obtained are combined sequently. Table 1 shows different methods used to
obtain the ranking of scores.
We have used four methods: MAX merges the n list and if a document is in several lists it will
provide the highest score. SUM carries out the average of the scores. The methods 3 and 4 are
as the previous ones but using normalization. This normalization is carried out subtracting the
score of each document RSVk from the minimum score of the list and dividing by max(RSVK ) −
min(RSVk ).
Obviously, this model improves and involves speed process. However in IR-n system archi-
tecture, this trouble does not increase the speed process of the system. This happens because
Number Method Formula
1 MAX max(RSVk )
2 SUM sum(RSVk )
3 MAX RSVnorm max((RSVk − min(RSVk ))/(max(RSVk ) − min(RSVk ))
4 SUM RSVnorm sum((RSVk − min(RSVk ))/(max(RSVk ) − min(RSVk ))
Table 1: Data fusion method
Language Collections TotalDocs Size SDAvg WDAvg WSAvg
English The Angeles Times 94 169477 579 MB 25 529 20
Glasgow Herald 95
French Le Monde 94/95 177452 487 MB 17 388 21
SDA French 94/95
Portuguese Público 94/95 210734 564 MB 18 433 23
Folha 94/95
Hungarian Magyar Hirlap 02 49530 105 MB 11 245 20
Bulgarian Standart 02 - Sega 02 69195 213 MB 246 157 18
Table 2: Data Collections
IR-n system produces a segmentacion of the documents in passages in the search time and the
calculation of similarity is carried out on structures load in memory.
3 Training
This section describes the training process which has been carried out in order to obtain the
best features to improve the performance of the system. Firstly, the collections and resources are
described. The following section explains the specific experiments which we have carried out.
3.1 Data Collections
This year our system has participated in the following tasks: Monolingual: English, French,
Portuguese, Bulgarian y Hungarian. Table 2 shows the characteristics of the collection which we
have worked.
• SDAvg is the average of sentences in each document.
• WDAvg is the average of words in each document.
• WSAvg is the average of words in each sentence.
3.2 Experiments
This year several tests have been carried out in order to establish the best similarity measure for
each language and to provide the value of input parameters in the system. We have evaluated
the following languages: English, French and Portuguese. We could not evaluated Bulgarian and
Hungarian languages because we did not have data of last years, as result of this, we have choosed
the similarity measures and the parameters for these languages comparing them with the rest
ones.
The aim of the experiments phase is set up the optimum value of the input parameters for
each collection. For training has been used the collections CLEF-2003 (English and French) and
CLEF-2004 (Portuguese). Query expansion techniques have also been used in all languages.
Language size P k1 b avg avgP
English 8 2 0.5 300 0.5083
French 9 1.2 0.3 300 0.5240
Portuguese (2004) 8 2 0.5 500 0.4741
Table 3: Training results in fixed passages system
Language size P k1 b avg exp expd expq avgP
English 8 1 0.5 500 4 5 10 0.5267
French 8 2 0.7 400 3 5 10 0.5488
Portuguese (2004) 10 2 0.5 500 3 10 10 0.5084
Table 4: Training results in fixed passages system with query expansion
3.2.1 Fixed size passages
It has been performed experiments for setting the size passages and the values of parameters in
okapi system which allow us to obtain the best results. As the table shows 3 the size passage is
the same for all languages ( 8 sentences), however in French is 9 sentences.
3.2.2 Fixed size passages with query expansion
On the one hand, experiments which are carried out with query expansion tried to fix the number
terms to add in the original query and the number of documents (passages) to take into account.
Furthermore, we have evaluated the use of different size passages. We have got the best results
with 10 terms in every test and it has been used the 5 or 10 passages more relevants depending
on the specific language. On the other hand, it can be appreciated that the size passage is 8 in
this case also.
As we check in the table 5 query expansion in fixed system allows us to improve scores between
3.6% and 7.2% according to the different languages.
3.2.3 Combined passages
Combined passages method consist on using the similarity values which are provided from different
size passages of the same document to obtain the document similarity. Because of that, it has
been defined three types of passages: small, medium and big passage. The number of sentences
which composes each passage is the following:
P 1 = (3, 4, 5, 6)
P 2 = (7, 8, 9, 10)
P 3 = (11, 12, 13, 14)
Experiments have been carried out by means of using one of each type. In this way is obtained
the similarity of each passage. Document similarity is got using one of the four method described
previously (see table 1).
Language avgP avgP with exp Dif
English 0.5083 0.5267 +3.6%
French 0.5240 0.5488 +4.7%
Portuguese (2004) 0.4741 0.5084 +7.2%
Table 5: Comparative results in fixed system
Language Model P1 P2 P3 avgP
English 1 3 8 11 0.5034
English 2 3 7 13 0.5182
English 3 5 10 12 0.4987
English 4 3 7 14 0.5139
French 1 5 8 12 0.5129
French 2 3 9 12 0.5208
French 3 3 8 12 0.5106
French 4 3 7 12 0.5163
Portuguese (2004) 1 6 8 13 0.4625
Portuguese (2004) 2 4 8 14 0.4792
Portuguese (2004) 3 4 8 14 0.4724
Portuguese (2004) 4 4 8 14 0.4788
Table 6: Training results in combined passages method
Language Model P1 P2 P3 avgP
English 1 5 9 11 0.5074
English 2 4 9 11 0.5344
English 3 6 7 11 0.5259
English 4 4 9 11 0.5317
French 1 5 9 13 0.5614
French 2 3 8 11 0.5599
French 3 3 9 11 0.5486
French 4 6 10 14 0.5576
Portuguese (2004) 1 4 8 11 0.4892
Portuguese (2004) 2 4 8 12 0.5089
Portuguese (2004) 3 4 8 12 0.5046
Portuguese (2004) 4 4 8 12 0.5085
Table 7: Training results in combined passages method with query expansion
The tests which get better results are showed in table 6. ’Model’ belongs to the combined
method used and the columns P1, P2 and P3 are size passages which are provided by the best
combined method. The columns P1, P2 and P3 are a small, medium and big passages respectively.
As we show the combined method, which provides the best results, is the method 2 (SUM
without normatization) for all languages.
We have proved that the results increases comparing with fixed system in all languages except
in French.
3.2.4 Combined passages with query expansion
We have carried out the same tests with query expansion and the results improves in all languages,
although the increase is not meaningful in Portuguese.
The best combined method for English and Portuguese carries on being the method 2 (SUM),
but in French is 1 (MAX) (see table 7).
As we check in table 5 combined passages system improves the results between 3.1% and 7.7%
according to each language.
Tables 9 and 10 compare both methods showing the best results obtained in each test.
Language avgP avgP with exp Dif
English 0.5182 0.5344 +3.1%
French 0.5208 0.5614 +7.7%
Portuguese (2004) 0.4792 0.5089 +6.1%
Table 8: Comparative results in combined system
Language avgP fixed avgP comb Dif
English 0.5083 0.5182 +1.9%
French 0.5240 0.5208 -0.6%
Portuguese (2004) 0.4741 0.4792 +1%
Table 9: Comparative: fixed system vs combined system without query expansion
4 Results at CLEF-2005
We have submitted three runs for each language in our paticipation at the CLEF-2005. The best
parameters, which provided the best results in the training, have been used in all cases. We did
not have any training data about Bulgarian and Hungarian languages, therefore we have used the
parameters of English language.
• IRn-xx-fexp is based on using the fixed size passages system which obtains the best results
in training. Query expansion techniques have been used in these runs.
• IRn-xx-vnexp Combined passages system has been used without applying query expansion
techniques.
• IRn-xx-vexp Combined passages system has been used with query expansion techniques.
Officials results for each run are showed in table 11. The model IRn-xx-vnexp is taken as a
reference. As other models which use query expansion techniques, our model also increases the
performance on the base system.
In this table 11 the two models with query expansion are compared. This one presents that
the percentage of improvement in the combined model is around 4% of increase avgP in every
language (except for Bulgarian).
As shown table 11, our results are above average in all languages appreciably, except for
Bulgarian that the results are below average.
5 Conclusions and Future Work
This paper has described PR model which uses similarity values of three different size passages for
each document in order to obtain the similarity of document regarding the question. This model
has allowed us to improve the results around 4% according to models which used only a fixed size
passage.
Language avgP fixed avgP comb Dif
English 0.5267 0.5344 +1.4%
French 0.5488 0.5614 +2.2%
Portuguese (2004) 0.5084 0.5089 +0.09%
Table 10: Comparative: fixed system vs combined system with query expansion
Language Run AvgP Dif
English IRn-en-vexp 41.88
IRn-en-fexp 41.08
IRn-en-vnexp 40.24
French CLEF Average 35.30
IRn-fr-vexp 35.90 +1.7%
IRn-fr-fexp 34.85
IRn-fr-vnexp 30.70
Portuguese CLEF Average 33.29
IRn-pt-vexp 36.03 +8.2%
IRn-pt-fexp 34.46
IRn-pt-vnexp 33.15
Hungarian CLEF Average 29.00
IRn-hu-vexp 31.74 +9.4%
IRn-hu-fexp 30.55
IRn-hu-vnexp 30.36
Bulgarian CLEF Average 22.00
IRn-bu-vexp 17.46
IRn-bu-fexp 17.58
IRn-bu-vnexp 17.87 -18.0%
Table 11: CLEF 2005 official results. Monolingual tasks
IR-n architecture allows us to realize this increase of steps in combined model without improv-
ing speed process considerably.
Lastly, we outline the future directions that we plan to undertake not only to improve this
model, but also to be applied it in Question Answering task.
6 Acknowledgements
This research has been partially funded by the Spanish Government under project CICyT number
TIC2003-07158-C04-01 and by the Valencia Government under project numbers GV04B-276 and
GV04B-268
References
[1] Aitao Chen and Fredric C. Gey. Combining query translation and document translation in
cross-language retrieval. In Carol Peters, Julio Gonzalo, Martin Braschler, and et al., edi-
tors, 4th Workshop of the Cross-Language Evaluation Forum, CLEF 2003, Lecture notes in
Computer Science, pages 108–121, Trondheim, Norway, 2003. Springer-Verlag.
[2] M. Kaskziel and J. Zobel. Passage retrieval revisited. In Proceedings of the 20th annual
International ACM Philadelphia SIGIR, pages 178–185, 1997.
[3] F. Llopis. IR-n: Un Sistema de Recuperación de Información Basado en Pasajes. PhD thesis,
University of Alicante, 2003.
[4] Fernando Llopis, Rafael Muñoz, Rafael M. Terol, and Elisa Noguera. Ir-n r2 : Using normalized
passages. In Carol Peters and Francesca Borri, editors, Cross Language Evaluation Forum:
Working Notes for the CLEF 2004 Workshop, pages 65–72, Pisa, Italy, 2004. IST-CNR.
[5] G. Salton. Automatic text processing: The transformation, analysis, and retrieval of informa-
tion by computer. 1989.
[6] Savoy J. Fusion of probabilistic models for effective monolingual retrieval. In Carol Peters, Julio
Gonzalo, Martin Braschler, and et al., editors, 4th Workshop of the Cross-Language Evaluation
Forum, CLEF 2003, Lecture notes in Computer Science, Trondheim, Norway, 2003. Springer-
Verlag.
[7] A. Singhal, C. Buckley, and M. Mitra. Pivoted document length normalization. In Proceedings
of the 19th Annual International ACM SIGIR Conference on Research and Developement in
Information Retrieval, Experimental Studies, pages 21–29, 1996.