Learning Restricted Regular Expressions with Interleaving from XML Data

Conceptual modeling in the era of Big Data and Artificial Intelligence: Research topics and introduction to the special issue

“…The papers presented in these two sections covered the topics of improving knowledge discovery processes or improving the description of techniques used in Machine Learning components or systems [36]. In the area of improving knowledge discovery, the topics included: deep learning techniques for incorporating reviews into recommendation systems [37]; mining rules with constants for knowledge base construction [83]; an approach for extracting structural relationships [51]; real-time event summarization from microblogging data [47]; natural language text classification [35]; and discovering regular expressions for schemas in XML documents [45].…”

An effective algorithm for learning single occurrence regular expressions with interleaving

“…The presence of a schema for XML documents has many advantages, such as for query processing and optimization, development of database applications, data integration and exchange [15,42,34,18]. However, many XML documents in practice are not accompanied by a (valid) schema [3,37,36,6,41,25], making schema inference an attractive research problem [2,5,7,17,22,43,13,30,32]. Studying schema inference also has several practical motivations.…”

“…In such cases the interleaving is necessary. Here we list the more recent efforts on RE(&) inference (see [13,40,43,30,32]). The aim of these approaches is to infer restricted subclasses of single occurrence RE(&), in which each symbol occurs at most once, starting from a positive set of words.…”

“…The content models of XML schemas are defined by regular expressions, and previous research has shown that the essential task in schema learning is inferring regular expressions from a set of given samples [9,2,5,7,17,22,43,13,30,32]. In fact, in some cases these learned regular expressions can directly be used as parts of the schema, and in other cases the inference of regular expressions is the most important component of the schema inference.…”

“…This means that no matter how many positive samples from the target language (i.e., the language to be learned) are provided, no algorithm can infer every target regular expression. Hence, researchers have turned to study subclasses of regular expressions [38,9,2,5,7,17,22,43,13,30,32].…”