Blueprint of a Cross-Lingual Web Retrieval Collection

Article excerpt

ABSTRACT: The world wide web is a natural setting for cross-lingual information retrieval; web content is essentially multilingual, and web searchers are often polyglots. Even though English has emerged as the lingua franca of the web, planning for a business trip or holiday usually involves digesting pages in a foreign language. The same holds for searching information about European culture, sports, economy, or politics. This paper discusses the blue-print of the WebCLEF track, a new evaluation activity addressing cross-lingual web retrieval within the Cross-Language Evaluation Forum in 2005.

Categories and Subject Descriptors

H.3 [Information Storage and Retrieval]: H.3.1 Content Analysis and Indexing; H3.3 Information Search and Retrieval; H3.4 Systems and Software; H3.7 Digital Libraries

General Terms

Measurement, Performance, Experimentation

Keywords: Information Retrieval, Cross-Language Information Retrieval

1. INTRODUCTION

The world wide web is a natural setting for cross-lingual information retrieval. This is particularly true in Europe: many European searches are essentially cross-lingual. For instance, when organizing to travel abroad for a business trip or a holiday, planning and booking usually involves digesting pages in foreign languages. Similarly, looking for information about European culture, sports, economy, or politics, usually requires making sense of web pages in several languages. A case in point is the current European Union, which has no less than 20 official languages.

The linguistic diversity of European content is "matched" by the fact that European searchers tend to be multilingual. Some Europeans are native speakers of multiple languages. Many Europeans have a broad knowledge of several foreign languages, and especially English functions as the lingua franca of the world wide web. Moreover, many Europeans have a passive understanding of even more languages.

The challenges of cross-lingual web retrieval will be addressed in WebCLEF [17], a new track at the Cross-Language Evaluation Forum [3, CLEF] in 2005. In this paper we provide a preliminary overview, discussing our view of the cross-lingual web retrieval task, the document collection used, EuroGOV, and the overall set-up of the WebCLEF track.

The remainder of the paper is organized as follows. In Section 2, we describe cross-lingual aspects of web retrieval in the European context. Section 3 discusses the problems involved, and outlines a test-suite for cross-lingual web retrieval. Then, in Section 4, we provide details of EuroGOV, a new web collection for cross-lingual web retrieval. Section 5 details how this collection will be used within the setting of the WebCLEF track at CLEF. Finally, in Section 6, we discuss our findings and draw some conclusions.

2. CROSS-LINGUAL WEB RETRIEVAL

In this section we discuss why the web is a natural habitat for cross-lingual information retrieval.

2.1 Multilingual Web Content and Users

The web is essentially multilingual. Although reliable statistics on web content and web usage are hard to come by, it is evident that the web is increasingly reflecting the linguistic diversity of the world's population. Let us first look at the web's content. Some indicative figures on the distribution of web content over languages are shown in Table I. (1) On the one hand, it is clear that English still functions as the lingua franca of the web. English is by far the most frequently used language. On the other hand, it is also clear that there is a substantial amount of non-English content on the web. The total amount of non-English pages is approaching that of pages in English. European languages other than English account for over a quarter of the global web content.

Let us now turn to the web's users. Table 2 gives, again, some indicative figures on the distribution of web users over languages. …