Visualization of Word Similarity Measurement for Messages in Sequence Diagram using Heatmap

The sequence diagrams emphasize time-based ordering and are very helpful for understanding the specifications and complex activities in each use case. This is important to analysts in compiling sequence-messages that create interactions between objects in the sequence diagram during the analysis stage. Later, during the design stage, these sequence-messages will be converted to operations in both the sequence diagram and class diagram. Reusing existing sequence messages is possible when analysts need to deal with complex sequences of messages in sequence diagrams. Analysts can apply reference fragments to reuse parts of one sequence diagram in another sequence diagram. In order to apply reusability in developing a sequence diagram, measure similarity among the sequence of messages using appropriate similarity models. Furthermore, it is more difficult for the analyst to trace the similarities of messages in a complicated sequence diagram based on the numerical representation of the results. Therefore, this study measured similarity between messages in each sequence diagram for the create, read, and update processes using Word2Vec, Bag of Words, and the TF-IDF web embedding tool. Then, the results of these similarity measurements will be visualized in heatmaps. A heatmap is a type of plot that is often used in multiple correspondence analysis. Based on the results, the comparison heatmaps showed the biggest differences between the cosine similarity and Euclidean distance using the Word2Vec method. This study will help analysts refine the most similar messages in the sequence diagram in the analysis stage.