An Approach for Maintaining Structural Uniformity of Multiword Expressions

In this work, a Minimum Edit Distance (MED)-based approach for lexical uniformity of a Multiword Expression (MWE) in Bengali text is described. MWE can take several different forms where there are blank spaces or hyphens ('-'), separating the constituent words. For Bengali MWEs, accuracy in recognition, classification, and sensory tagging tasks decreases by 5% because of the diverse architectures of MWEs. Ensuring structural consistency is a crucial preprocessing effort to construct a reliable system for the syntactic and semantic processing of Bengali MWEs. An F1-Score of 90.58% is measured as the algorithm's performance.