Semantic Similarity Match for Data Quality

Data quality is a critical aspect of applications that support business operations. Often entities are represented more than once in data repositories. Since duplicate records do not share a common key, they are hard to detect. Duplicate detection over text is usually performed using lexical approaches, which do not capture text sense. The difficulties increase when the duplicate detection must be performed using the text sense. This work presents a semantic similarity approach, based on a text sense matching mechanism, that performs the detection of text units which are similar in sense. The goal of the proposed semantic similarity approach is therefore to perform the duplicate detection task in a data quality process.