Abstract:New technologies allow to store vast amount of data about users interaction. From those data the social network can be created. Additionally, because usually also time and dates of this activities are stored, the dynamic of such network can be analysed by splitting it into many timeframes representing the state of the network during specific period of time. One of the most interesting issue is group evolution over time. To track group evolution the GED method can be used. However, choice of the timeframe type and length might have great influence on the method results. Therefore, in this paper, the influence of timeframe type as well as timeframe length on the GED method results is extensively analysed.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: **The influence of the type and size of time windows on group evolution discovery in dynamic social networks**.
Specifically, with the development of new communication technologies, a large amount of user interaction data can be stored and social networks can be constructed from it. Since this data usually contains information about the time and date of activities, its dynamic changes can be analyzed by dividing the network into multiple time windows. One important research question is **the evolution of groups (or communities) over time**. In order to track group evolution, a method called GED (Group Evolution Discovery) can be used. However, choosing different types and lengths of time windows may significantly affect the results of the GED method.
Therefore, the main purpose of this paper is to experimentally analyze the influence of different time window types (such as non - overlapping, overlapping, and increasing) and different window sizes on the results of the GED method, in order to find the optimal time window settings and thus more accurately track and understand the group evolution process in social networks.
### Key issues
1. **Selection of time window types**: including non - overlapping (disjoint), overlapping (overlapping), and increasing (increasing) time windows.
2. **Selection of time window sizes**: Different - sized time windows may affect the detection results of group evolution.
3. **Parameter adjustment of the GED method**: The influence of the selection of α and β parameters on the results.
### Experimental results
- **Non - overlapping time windows**: Since there is no overlap between time windows, there are fewer group evolution events (such as growth, shrinkage, splitting, merging, etc.), and mainly the formation and dissolution events of groups are detected.
- **Overlapping time windows**: Increasing the overlap between time windows makes the user interactions in adjacent time windows have more overlap, so that the group evolution events can be better captured.
- **Increasing time windows**: Each subsequent time window contains the relationships and nodes of all previous time windows, so groups last longer, which is suitable for studying so - called "persistent groups".
### Conclusions
For rapidly changing social networks, **overlapping time windows** are the best choice because they can provide more group evolution events. And **increasing time windows** are more suitable for studying persistent groups. In addition, the type and size of time windows can be adjusted as an additional parameter of the GED method to make it more flexible and practical.
Through this research, the author provides guidance for future researchers on how to choose appropriate time window types and sizes in order to more effectively analyze group evolution in social networks.