Recently, the Solutions Development department at Audimation took on a project with a major client who needed to analyze data from the Asia Pacific area for red flag key words in order to comply with the Foreign Corrupt Practices Act (FCPA). The client needed to monitor the data entered into transaction descriptions to detect evidence of bribery and other FCPA violations using their travel and entertainment, accounts payable, and accounts receivable data. On the surface, it looked like the sort of work the members of the team had done many times before.
The catch? The data wasn’t in English.
Faced with the difficult task of trying to analyze foreign language text, Solutions Development turned to translation software to get the job done. The script that was developed for this client’s Korean and Chinese data exports IDEA databases to text where they're immediately picked up by the translator engine. The translator engine for that particular language produces its own text file of the data that it has translated to English. That data is then imported back into IDEA where it can be analyzed for red flag key words. The translator, which has an engine for translating 30 different languages, runs in the background and performs its work automatically.
This technique is highly effective because it analyzes slang specific to each language translated. For instance, the term “chai” in East Africa is slang for bribe. In North America, it simply means a type of tea. Because languages vary in the terms used to describe acts like bribery, this method of translating and analyzing text data uses a master red flag keyword list in English for each language translated, allowing for great flexibility for proper translation and analysis of culture-specific terms.
Potential applications go beyond just detecting bribery. Unauthorized vendors, travel and expense fraud, or unauthorized purchases of any kind are among the many potential uses Solutions Development has identified for this particular project and its powerful text analysis capabilities.
With this solution, Audimation has shown that language doesn’t have to be a barrier in data analysis.
Number of views (982)/Comments (0)