Inference of location where each textual message was posted

A system for inferring a geographical location of a user based on a textual message posted by the user via a mobile device, the system comprising:

• a processor;
• and
• a memory storing a program, which, when executed on the processor, performs a method comprising:
• partitioning an area identified from the textual message into a plurality of sub-areas each corresponding to a grid on an electronic map;
• and
• determining the geographical location of the user based on one or more training feature vectors including one or more scores related to keyword importance of one or more respective training keywords extracted from one or more training textual messages posted by one or more training users via one or more training mobile devices and transition probabilities for respective ones of the plurality of sub-areas each corresponding to a probability that a training user is a distance away from its respective sub-area after an amount of time, the transition probability for a first sub-area being calculated based on an average velocity, the average velocity being calculated over a plurality of velocities between the first sub-area and a plurality of other sub-areas;
• wherein one or more training textual messages are associated with respective geographical locations corresponding to sub-areas where the one or more training textual messages were posted by the one or more training mobile devices as determined from publicly available data, and wherein determining the geographical location further includes:
• extracting one or more keywords from the textual message;
• constructing one or more feature vectors including one or more scores related to keyword importance of respective ones of the one or more keywords;
• computing, using the one or more feature vectors and the one or more training feature vectors, a plurality of scores each corresponding to a respective candidate sub-area associated with a candidate geographical location where the textual message was posted;
• and
• computing, based on the plurality of scores and the transition probabilities, time-series of location information to infer the geographical location of the user.

A computer program product for inferring a geographical location of a user based on a textual message posted by the user via a mobile device, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:

• partitioning an area identified from the textual message into a plurality of sub-areas each corresponding to a grid on an electronic map;
• and
• determining the geographical location of the user based on one or more training feature vectors including one or more scores related to keyword importance of one or more respective training keywords extracted from one or more training textual messages posted by one or more training users via one or more training mobile devices and transition probabilities for respective ones of the plurality of sub-areas each corresponding to a probability that a training user is a distance away from its respective sub-area after an amount of time, the transition probability for a first sub-area being calculated based on an average velocity, the average velocity being calculated over a plurality of velocities between the first sub-area and a plurality of other sub-areas;
• wherein one or more training textual messages are associated with respective geographical locations corresponding to sub-areas where the one or more training textual messages were posted by the one or more training mobile devices as determined from publicly available data, and wherein determining the geographical location further includes:
• extracting one or more keywords from the textual message;
• constructing one or more feature vectors including one or more scores related to keyword importance of respective ones of the one or more keywords;
• computing, using the one or more feature vectors and the one or more training feature vectors, a plurality of scores each corresponding to a respective candidate sub-area associated with a candidate geographical location where the textual message was posted;
• and
• computing, based on the plurality of scores and the transition probabilities, time-series of location information to infer the geographical location of the user.