CenSEARCHip

Most people are aware that the major search engines abide with censorship restrictions in some countries in order to be able to do business there. China is the prime example, but there are others - France and Germany, for instance. A new visual tool created by a doctoral student at the University of Indiana gives an interesting comparison between the search results for different countries' versions of search engines.

The tool is called CenSEARCHip, and there is a helpful summary of it on First Monday. For image searches, the tool returns thumbnails that can be visually compared by the user. For text searches, the results come in the form of a "tag cloud"-like representation, where the terms that appear much more frequently for one search engine than the other are in larger font.

The visual display of the comparison is a distinctive feature of this tool compared to others that have been developed. The authors are aware that some challenges in interpretation still exist, such as being able to know which differences are the result of censorship and which are not.

I tried comparing Google Image Search for China and the United States with the search term "Nanking." The results were substantially similar. This contrasts with the example of "Tianenmen Square" given in the First Monday paper above, where the results were quite different. It made me wish that Japan were a comparison option. Next I tried "e-waste," a term of little political importance, but one where China may be associated with negative terms. The China/US comparison produced few interesting differences.

"Falun gong" produced some notable differences - in particular, the greater popularity of the terms "mentally ill" and "obsessed" in the Chinese version. One helpful modification of the CenSEARCHip tool would be the ability to recognize groups of words together - "mentally" and "ill" actually appeared as separate terms in the results.

I agree with the authors overall that the tool has some merit, and also could be improved. I appreciate that the interface and displayed results are simple and intuitive. It would be interesting to probe how different users interpret the results in order to understand how people find meaning in image and tag-cloud comparisons; I am not sure we know that.