电子学报 ›› 2022, Vol. 50 ›› Issue (11): 2568-2574.DOI: 10.12263/DZXB.20211164

• 学术论文 • 上一篇    下一篇

旁观者视角下粒的多层次描述

智慧来1, 张丽1, 李金海2   

  1. 1.河南理工大学计算机科学与技术学院, 河南 焦作 454003
    2.昆明理工大学数据科学研究中心, 云南 昆明 650500
  • 收稿日期:2021-08-28 修回日期:2021-11-25 出版日期:2022-11-25
    • 通讯作者:
    • 李金海
    • 作者简介:
    • 智慧来 男,1981年3月出生于河南省洛阳市.2010年毕业于上海大学计算机工程与科学学院.现为河南理工大学计算机科学与技术学院副教授、硕士生导师.主要研究方向为形式概念分析、粗糙集、粒计算等.E‑mail: zhihuilai@126.com
      张 丽 女,1995年4月出生于河南省焦作市.现为河南理工大学计算机科学与技术学院硕士研究生.主要研究方向为形式概念分析、粒计算等.E‑mail: liizhang13@163.com
      李金海(通讯作者) 男,1984年1月出生于江西省上饶市.2012年毕业于西安交通大学数学与统计学院.现为昆明理工大学数据科学研究中心教授、博士生导师.主要研究方向为概念格、粗糙集、模糊集、粒计算、认知计算等.
    • 基金资助:
    • 国家自然科学基金 (11971211); 河南省高校基本科研业务费专项 (NSFRF210318)

Multi-Level Description of Granules From an Outsider’s Perspective

ZHI Hui-lai1, ZHANG Li1, LI Jin-hai2   

  1. 1.School of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, Henan 454003, China
    2.Data Science Research Center, Kunming University of Science and Technology, Kunming, Yunnan 650500, China
  • Received:2021-08-28 Revised:2021-11-25 Online:2022-11-25 Published:2022-11-19
    • Corresponding author:
    • LI Jin-hai

摘要:

不确定性在大数据中普遍存在.现有的研究极少从外部视角研究数据的不确定性,特别是模糊数据,经常无意识地假定获得的原始数据是完全真实可靠的,不存在任何偏差.然而,鉴于数据获取的主体普遍存在着不确定性,在大多数情况下这种做法是不可取的.为此,本文基于模糊概念格提出旁观者视角下粒的多层次描述.该方法基本思想是通过单个元素隶属度找出模糊形式背景中所有的原子粒,利用原子粒的自由组合来判定一个目标粒是否为可定义粒,并由原子粒的描述获取可定义粒的描述.接着,借助粗糙集上下近似思想给出一种不可定义粒的近似描述方法,它的主要思想是用可定义粒近似刻画不可定义粒.其中,查全率和查准率是为不可定义粒选择最合适近似描述的两个重要评价指标,而近似描述的质量则通过描述精度进行度量.最后,通过商品评价的实验分析说明了所提多层次数据分析方法的必要性与可行性,同时表明模糊概念格的真值越丰富得到的多层次粒描述结果越好.

关键词: 不确定性, 模糊集, 粒计算, 概念格, 粒描述, 模糊概念, 多层次数据分析

Abstract:

Uncertainty is pervasive in big data. In the existing researches, the uncertainty of data was seldom concerned from an external perspective, especially for the fuzzy data, and it was often unconsciously assumed that the original obtained data is completely truthful and reliable without any bias. However, due to the ubiquitous uncertainty of the subject in data acquisition, this kind of assumptions are not reasonable in most cases. In order to solve this problem, a multi-level description method of granules is proposed from an outsider’s perspective based on fuzzy concept lattice. The basic idea of this method is to find out all the atomic granules in the fuzzy formal context by using the degree of membership of single element, and further determine whether a target granule is a definable granule by a free combination of the atomic granules, which can be used to obtain the description of a definable granule by the descriptions of atomic granules. After that, an approximate method to describe indefinable granules is proposed based on the ideas of the lower and upper approximations which are from the rough set theory, and the main idea of this method is to describe the indefinable granules approximately by means of definable granules, where recall and precision are two important evaluation criteria to select the most appropriate approximate descriptions for the indefinable granules, while the quality of the approximate descriptions is measured by description accuracy. Finally, the necessity and feasibility of the proposed multi-level data analysis methods are illustrated by an experimental analysis of commodity evaluation, and at the same time, the bigger the number of membership degrees in the fuzzy concept lattice, the better the obtained results of multi-level granule description.

Key words: uncertainty, fuzzy set, granular computing, concept lattice, granule description, fuzzy concept, multi-level data analysis

中图分类号: