关键词: 设计概算, 定额绑定, 自然语言处理, 信息检索, 词义分析

Abstract: Budget estimate quotas and mechanical quotas are the basis of construction resource analysis and engineering budget estimate compilation. Traditional manual binding of the process and quota is inefficient, cumbersome and error-prone. Based on establishing a quota paradigm database, this paper presents a new intelligent binding method for budget estimate quotas of construction process by using information retrieval and natural language processing technologies. For the budget estimate quotas with positive and negative relevance, a fuzzy retrieval model is developed using the morpheme segmentation of text keywords and combining Word2vec semantic analysis with inverse document frequency weights. And we construct a priori cascading search method for budget estimate quotas in response to the characteristics of hierarchical search of the quotas. Experiments show that the retrieval model achieves a recall precision of 92.29% for budget estimate quotas and 97.28% for construction machinery hourly rate quotas, and the search method reduces design man-hours significantly. Application of this binding method will be a strong support to intelligent designs in hydropower engineering.

Key words: design budget estimate, quota binding, natural language processing, information retrieval, semantic analysis