Detecting Mobile Misinformation by Using Mobile Text Mining

Smartphones are indispensable devices for people in these days, and tens or even hundreds of messages are sent to each device every day. All kinds of information can be found from the messages such as news, greetings from family members or friends, advertisements, promotions, weather reports, etc. People are overwhelmed by the sheer amount of information and they spend much time to sort out a way to find relevant information from the messages. Even worse is some messages give false or fake information, and mislead the viewers consequently. This research tries to automatically classify the messages received by a device into several categories such as advertisements, news, messages from family or friends, announcements, etc. even misinformation by using various mobile text/data mining technologies. The proposed method takes the following steps:
  1. Mobile data collection: Collect all data that can be found such as user id, text messages, time and date received, hyperlinks, hypertext, etc.
  2. Mobile data preparation: Not all collected data is useful or can be used like video or audio clips, and has to be removed. In addition, the collected data such as times and dates may need to be changed to a standard format.
  3. Mobile data indexing and storage: The prepared data will be saved into a database and indexed for fast searching and retrieval.
  4. System training: Text/data mining technologies (like artificial neural networks or data classification) will be used to classify the messages. The first step is to train the system by using the data we collected for a short period of time. For example, the data we collect may include sender’s id, time and date of the data received, the keywords or phrases like coronavirus, COVID-19, cured, free, credit card number, etc. In addition, we also need to tell the system to what categories like advertisement and false information this message belongs.
  5. Applications: After the system is trained, it will be put to use. At the same time, the system will keep learning or may be re-trained to become better.
People are flooded by a plethora of mobile data every day. Some of the data is relevant or important, but some of it is irrelevant or even harmful. This research tries to mitigate this problem by using the technologies of mobile text/data mining, so users will benefit by the resourceful messages.

Keywords: Mobile computing, security and privacy, mobile data management, mobile text/data mining, mobile misinformation