DCNL: Disclosure Control of Natural Language Information to Enable Secure and Enjoyable E-Communications

Natural language communications using social networking and blogging services can result in the undesired revelation of private information. Existing disclosure control is tedious and error-prone because the user must set the disclosure level manually and must reconsider the level every time a new text is to be uploaded. This can lead to the revelation of private information or reduced enjoyment of the communication due to either disclosing too much text or hiding text that is meant to be shared. To solve these problems, we are developing a new disclosure control mechanism called DCNL or disclosure control of natural language information. DCNL automatically checks texts uploaded to social networking services or blog pages, detects words that might reveal private information, and warns the user about them. The granularity of DCNL is not the text but the words in the text. Consequently, it is not tiresome for the user and balances the protection of privacy with the enjoyment of communications. DCNL checks not only words that directly represent private information but also those that indirectly suggest it. Combinations of words are also checked. Analysis of the co-occurrence between words and reachability analysis with a search engine are used to infer what words imply what information.