Abstract: Nowadays all kinds of social platforms provide various
recommendation services, greatly enriching people's life. In the condition
that social platforms have become indispensable tools, people have more to
except that the social platforms can provide more convenient and efficient
services for life and work. With three dimensions of context: time, place
and activity, this paper presents one new real-time recommendation
methodology based on user's schedule. It transforms the schedule-based user
recommendations into text clustering using K-means algorithm. Due to the
predefined limit of value K in traditional K-means algorithm, the K-means
algorithm is improved and an application of improved K-means algorithm in
the recommendations model of schedule-based users is introduced in this
paper. The results of relevant experiments and analysis indicate that the
system improved the recall and precision than traditional one and has
practical application value.
Keywords: K-means, text clustering, schedule-based user recommendation, three-dimensional context
