Multipurpose semantic analysis of construction text documentation

Construction contracts are crucial documents as they outline the agreement. In case any vagueness is present in contract sentences, the interpretation of the sentences will vary between different parties of the contract. Such a case would cause a dispute and eventually adversely affect the construction project's success. Additionally, multiple parties that sign the contract and various departments/disciplines within a party need to be aware of the content of these contracts. All these departments review and apply the contract conditions only from their perspective. An additional effort is required for the coordination of such different departments. The above examples are related to the practice of contract administration and interpretation processes of contract documents. The conventional approach for these processes is manual, prone to errors, requires expert involvement, and is time-consuming. Automation efforts for such manual processes would remove such deficiencies, as it will provide a certain level of standardization. A significant part of these contracts is text documents, which provide an unstructured nature of data. Recent advancements in the artificial intelligence field allow tackling the hardship of studying and analyzing unstructured data. Thus, this study proposes an approach that comprises a sequential application of natural language processing and supervised machine learning applications to desired automation. The proposed methodology reduces time spent in coordination efforts of contract review, also reliably and accurately predicts (i) classification of contract sentences for departmental relevance, (ii) vagueness presence in contract sentences, and also removes the dependence on expert participation.