Cross-Language Image Retrieval

Cross-language information retrieval (CLIR) has been receiving a lot of attention in the past two decades, mainly due to the popularisation of the Internet after the mid-1990s. Ever since, researchers focusing on the area of information retrieval have been running a host of experiments, trying out upgraded and newly developed CLIR systems with an aim to facilitate retrieval of documents that are written in language other than the language of the query. Though still far from being perfect, modern CLIR systems are exceptionally reliable if used properly.

CLIR Research Almost Exclusively Focused on Retrieving Textual Information

The same strategy that is used for retrieving textual information was for a long time also considered the best method for retrieving images which are normally accessed with the aid of the short text that is used to describe their contents. Since the user normally understands what the image is about, even if they don’t understand the language used to describe it, much of the research on the question of cross-language image retrieval was centred around the idea that the retrieval system can be considered successful as long as it brings the user to relevant information. According to this theory, the user will assess the image’s relevance based on its content rather than textual description.

Exceptions that Confirm the Rule

In most cases, the user doesn’t need to understand the caption text or the text that is used to describe the image. For example, no matter which language(s) they speak, everyone understands a picture of a bird, dog, landscape, cloud, etc. However, if the image is supposed to represent a particular bird species or dog breed for instance, the user may not be able to evaluate its relevance if they don’t understand the caption text. This means that the assumption that the user can assess the image’s relevance based on its content is in some instances incorrect. Therefore, cross-language image retrieval requires an approach that addresses the issue of relevance from all aspects, including the question of caption text.

Cross-Language Image Retrieval Also Presenting a Major Opportunity

Cross-language image retrieval is still presenting a major challenge for the IR researchers, however, it also presents a major opportunity. For example, an effective CLIR system that will address all or at least most of the “flaws” of the existing system and at the same time be easy to use, has a huge potential in the image collection market.