For the Phase 2 development project of the ACT Ai database of political office holders and senior civil servants, initiated by the Anti-Corruption Organization (Thailand), our Boonme Lab team gathered information from various websites to create a database. We also designed a system for filling out forms in PDF format to digitize the data for better usability.

The challenges included the paper forms having multiple versions even for the same type of document, inserting additional pages, incomplete documents, users not filling out forms according to the specified headings, and inconsistent formatting—such as entering numbers or decimals instead of fractions. These issues required our team and the Hand team to constantly discuss and find the best solutions because the users filling out the forms are not the data owners but readers who are entering the information as instructed. This made decision-making more difficult due to concerns about data accuracy.

Our data entry system allows contractors to fill out documents as assigned by the admin. Once the forms are completed, the system compares the data with its reference set, meaning each document set has three comparison sets filled out by three different contractors. If all three entries match, the system stores the document in the database for future use. If discrepancies are found, an admin will review the document again. Once verified, the data will also be stored in the database.

Designing the database and entry interface involved frequent adjustments. The more data we entered, the more hidden issues we uncovered, as the numerous documents filled out by contractors inevitably contained errors and unexpected surprises.

ACT AI Database Development