融合内容分析和关联分析的短生命周期体验品需求特征模式挖掘方法研究

挖掘特定产品的需求模式无法从整体掌握该类产品的市场特征；短生命周期体验品因缺乏历史销售数据，并且销售总量波动性极大，尤其需要从整体掌握销售总量与产品属性间关系的需求特征规律，但又难以挖掘，亟待提出适用于该类产品的需求特征模式挖掘方法。基于按销售总量分区后各区的需求特征的规律性，提出了一种按销售总量分区、以已有产品介绍集和销售总量为源信息、适用于新产品开发前使用、融合内容分析和关联分析的短生命周期体验品需求特征模式挖掘方法。该方法包括基于内容分析法的产品属性挖掘方法和基于关联分析的产品属性关系模式挖掘方法。前者可以得到较全面的产品属性；后者能够构建不同销售总量区间内产品集的属性关系模式，得到各区间的产品属性关系网，获得高销售总量区间具备，但中、低区间不具备的属性关系模式，从而获得需求特征模式。通过不断更新产品介绍集和销售总量并迭代挖掘，该方法能够动态挖掘需求特征模式。最后利用2013至2016年国产犯罪和爱情类电影数据验证了该方法的可行性，并得到了这两类电影的产品属性及近年的需求特征模式，可用于指导这两类电影的创作。

A Demand Feature Pattern Mining Method for Short Life Cycle Experiential Product by Integrating Content Analysis and Association Analysis

Specific product demand patterns fail to shed light on market characteristics of such a kind of products. For short life cycle experiential products, a lack of sales data and immense demand volatility make it necessary to have a thorough understanding of demand feature of the kind of products in a holistic way. However, the demand feature is difficult to identify. To address the issue, it is imperative to propose a method for mining demand feature patterns suitable for short life cycle experiential products.Immense demand volatility of a kind of products make it difficult to mine such a kind of products' relationships between product attributes and its total sales volume. After segregating a kind of products into several groups according to their total sales volume, fluctuation of each group's total sales volume becomes relatively small, and thus demand feature of each group with a narrow sales volume interval may be more regular, so it may be easy to mine the demand feature of each group. Based on this, a product demand feature pattern mining method is proposed for short life cycle experiential products by grouping such a kind of products according to their total sales volume.The method takes updated product introduction set and sales volume of existing short life cycle experiential products as source information. By integrating content analysis and association analysis, the method consists of two processes:(1) product attribute mining process through content analysis,(2) product attribute pattern mining process for each group through association analysis. Through content analysis at the first process, product attribute database is established; relatively comprehensive product attributes are obtained; product attribute table is set up. At the second process, through product attribute association analysis and through visual construction of demand feature resulted from the analysis, product attribute relationship pattern which is possessed by product set grouped in the high sales volume interval but not in the middle and low sales volume interval is mined out, and thus a dynamic demand feature pattern reflecting market demand trend is obtained.On the basis of data of 65 crime movies and 190 romantic movies shown in China from 2013 to 2016, the demand feature pattern mining method is verified, and demand feature pattern of crime movies and that of romantic movies in recent years are obtained. The patterns may be used to guide creation of these two types of movies.The demand feature pattern mining method is not only suitable for short life cycle products, but also suitable for long life cycle products with product introduction, because the method only uses product introduction and sales volume as source information. The method can be applied before products get developed, and thus demand feature pattern resulted from the method may be used to guide design of new products and may help to reduce investment risks of new products.