WO2018068603A1 - 基于大数据技术的供应链管理决策支持系统 - Google Patents

基于大数据技术的供应链管理决策支持系统 Download PDF

Info

Publication number
WO2018068603A1
WO2018068603A1 PCT/CN2017/101202 CN2017101202W WO2018068603A1 WO 2018068603 A1 WO2018068603 A1 WO 2018068603A1 CN 2017101202 W CN2017101202 W CN 2017101202W WO 2018068603 A1 WO2018068603 A1 WO 2018068603A1
Authority
WO
WIPO (PCT)
Prior art keywords
text
module
supply chain
chain management
decision support
Prior art date
Application number
PCT/CN2017/101202
Other languages
English (en)
French (fr)
Inventor
倪伟定
杜坚民
蔡日星
蔡一帆
Original Assignee
香港纺织及成衣研发中心有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 香港纺织及成衣研发中心有限公司 filed Critical 香港纺织及成衣研发中心有限公司
Publication of WO2018068603A1 publication Critical patent/WO2018068603A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
    • G06Q10/0875—Itemisation or classification of parts, supplies or services, e.g. bill of materials

Definitions

  • the invention relates to the field of information processing, in particular to a supply chain management decision support system based on big data technology.
  • EDI electronic data interchange systems
  • Supply chain managers make high-quality business or management decisions for management, planning, and operational tasks based primarily on supply chain members' transaction data and internal system data.
  • useful data from other sources especially data such as Twitter, blogs, news, emails, and Facebook, are not analyzed and involved in decision making.
  • the present invention proposes a supply chain management decision support system for data such as Twitter, blog, news, mail, and Facebook that are not analyzed and involved in supply chain management decision making.
  • the invention provides a supply chain management decision support system, comprising:
  • An information acquisition module configured to extract data from a big data source, convert the format thereof, and send the data to the analysis processing module;
  • An analysis processing module configured to search for useful information from the data sent by the information acquisition module, and integrate and analyze the same to provide an analysis result
  • a visualization module for displaying analysis results
  • Support modules including knowledge base, textile taxonomy database and access control sub-module; knowledge base Used to provide historical data for the information acquisition module and the analysis processing module for completing the analysis task; the textile taxonomy database is used to provide a textile classification method for the information acquisition module and the analysis processing module for completing the analysis task; access control The module is used to assign rights to the account on the login supply chain management decision support system.
  • the information acquisition module includes:
  • a content filter for filtering out noise data in a big data source
  • a text preprocessing unit for parsing data filtered by the content filter to obtain text
  • a named entity identification unit for locating and classifying words in the text according to predefined categories
  • a domain classification unit for dividing text into different fields according to predefined rules
  • a content scoring unit that scores the relevance of the text to the textile and fashion industries.
  • the indexing unit is configured to calculate the weight of the plurality of words in the text by using the TF-IDF algorithm, and take the preset number of words with the largest weight as the keyword.
  • the analysis processing module includes a content mining sub-module, a supply chain management analysis engine, and a decision support sub-module;
  • the content mining sub-module is used to process text and find useful information from it;
  • the supply chain management analysis engine is used to analyze the data sent by the information acquisition module to provide a supply chain management analysis report
  • the decision support sub-module is used to store the decision rules and solutions corresponding to the supply chain management analysis report respectively; and is also used to provide a corresponding solution to the supply chain management analysis report based on the corresponding decision rules.
  • the content mining sub-module is used for step-by-sentence scanning of the title and content of the text, and based on the scoring scheme, giving a corresponding basic score to the emotional words in the text sentence, and also for the emotional words.
  • Prefix words are checked, and different additive components are given according to the degree of change of the emotional words according to the prefix words, and then the product of the added component numbers and the base scores is calculated, which is recorded as the score of the sentence; and is also used to put all the sentences in the text.
  • the scores are combined and normalized to calculate The score of the text;
  • the content mining sub-module is further configured to arrange the text related to the product or the marketing activity according to the score of the text; the visualization module is configured to display the text related to the product or the marketing activity, which is arranged according to the score of the text.
  • the content mining sub-module is further configured to compare the score of the text with the preset positive emotion threshold and the preset negative emotion threshold; if the score of the text is greater than the preset positive emotion threshold, The text has a positive emotion; if the score of the text is greater than the preset negative emotion threshold and less than the preset positive emotion threshold, the text has a neutral mood; if the score of the text is less than the preset negative emotion threshold, the text has a negative emotion;
  • the Content Mining sub-module is also used to calculate the subjectivity of text related to a product or marketing campaign:
  • S is the subjectivity of the text associated with the product or marketing campaign
  • PA is the number of texts with positive emotions
  • NA is the number of texts with negative emotions
  • ZA is the number of texts with neutral emotions
  • a visualization module is used to display the subjectivity.
  • the content mining sub-module is also used to calculate the polarity of text related to the product or marketing activity:
  • P is the polarity of the text associated with the product or marketing campaign
  • PA is the number of texts with positive emotions
  • NA is the number of texts with negative emotions
  • a visualization module is used to display the polarity.
  • the big data technology-based supply chain management decision support system of the present invention can provide the following supply chain management services: 1) The supply chain management decision support system can reorder the service time, reorder content, and reorder based on the product positioning analysis result. The location and the number of reordering suggestions; 2) The supply chain management decision support system can remind the user to take appropriate actions based on the customer sentiment report when the user's extreme emotions are discovered. 3) The supply chain management decision support system can also be reported in the cost analysis The report shows that the cost of using the inefficiency can show users which cost is inefficient, such as the cost of raw materials is higher than the previous contract standard; in order to find opportunities to save costs, users will find ways to save costs, thereby improving cost efficiency. .
  • FIG. 1 is a schematic diagram of functional modules of a supply chain management decision support system according to an embodiment of the present invention
  • FIG. 2 is a block diagram showing a program code of a scoring scheme of a supply chain management decision support system according to an embodiment of the present invention
  • Figure 3 shows a program code diagram of a clustering hierarchical clustering
  • FIG. 4 is a data classification diagram obtained by the decision support sub-module shown in FIG. 1;
  • Figure 5 shows a flow chart of product demand forecast adjustments.
  • the supply chain management decision support system based on big data technology includes: an information acquisition module (for information extraction, conversion and loading), an analysis processing module, a support module, and a visualization module.
  • the information acquisition module is used to extract data from a big data source and convert its format to an analysis processing module; the big data source may include other databases (such as ERP and POS), supply chain members, and the Internet.
  • the data acquired by the information acquisition module from the big data source includes structured data and unstructured data.
  • structured data is defined as data that can be logically expressed using a database two-dimensional logical table, such as row data; non-structured data is defined as data that is inconvenient to logically express the implementation using a database two-dimensional logical table, such as Office documents, text, images, subsets of standard universal markup languages, HTML, audio, and video in all formats.
  • the information acquisition module can perform conversion processes such as standardization, verification, encoding, merging, and splitting to implement format conversion.
  • conversion processes such as standardization, verification, encoding, merging, and splitting to implement format conversion.
  • unstructured data the present invention focuses primarily on text, and the acquired text content is primarily in Chinese or English.
  • the information obtaining module includes:
  • a content filter that filters out noise data from large data sources may contain low-quality content such as misspellings, unrealizable content (such as rum), and malicious content, which need to be filtered out before being parsed.
  • filtering out noise data can be achieved by examining data sources (such as checking whether the data source is a trusted website, etc.), checking checks based on other texts, and using customary knowledge and linguistic rules.
  • a text preprocessing unit for parsing data filtered by the content filter to obtain text; specifically, the text preprocessing unit performs cleaning and parsing tasks, such as removing, segmenting, finding stems, or Lexical restoration; the text preprocessing unit is also used to store text in a structure suitable for analysis, such as a vector space model.
  • An indexing unit for finding and determining a keyword in the text searches for a keyword by weighting the word in the text.
  • the weight is calculated by the TF-IDF algorithm, ie
  • w i,j represents the weight of the word i in the text j
  • Tf i,j represents the word frequency of the word i in the text j
  • Idf i represents the inverse document frequency of the word i
  • N the total number of texts
  • n i represents the number of texts with the word i.
  • the indexing unit takes the preset number of words with the largest weight as the keyword.
  • index unit can represent the text j as a corresponding vector.
  • w in,j represents the weight of the word in in the text j.
  • unstructured data can be converted into structured data.
  • Named entity recognition unit for locating and dividing words in text according to predefined categories Class; for example, the named entity identification unit may classify words according to person name, organization name, geographical name, product name or product category, and emotional expression, and identify corresponding interpretations for these words. For example, for the phrase "The new sundress is available in Uniqlo in 2015”, after the name entity recognition unit is processed, it will become "2015 [time] sundress [product category] new has been in Uniqlo [company] Sale". Among them, it can be seen that time and company will be identified, and the word "sun skirt" will be marked by product category based on the textile industry classification.
  • a domain classification unit that divides text into different realms according to predefined rules; for example, based on predefined rules, text is divided into social media (including Weibo, Facebook, and Twitter), news, and e-commerce.
  • a content scoring unit that scores the relevance of the text to the textile and fashion industries.
  • the scoring process is implemented based on predefined rules. For example, if there is a textile-related keyword in the text, the text will be evaluated as a higher score; otherwise, there is a "unwanted" definition in the predefined rule. The text of the keyword will be evaluated for a lower score. The content scoring unit then arranges the text according to the score for further analysis.
  • An analysis processing module configured to search for useful information from the data sent by the information acquisition module, and integrate and analyze the same, thereby providing an analysis result;
  • the analysis processing module includes three sub-modules: a content mining sub-module, and a supply chain management (That is, Supply Chain Management) analysis engine and decision support sub-module.
  • the content mining sub-module is used to mine useful information in the unstructured data, specifically, the data content, the competitive advantage and the opinion are determined from the data sent by the information acquiring module.
  • the content mining sub-module can be used to process text and find useful information therefrom.
  • the content mining sub-module can determine customer sentiment, content clustering, and competitive advantage through an algorithm.
  • Emotional analysis involves the mining of ideas and analysis of customers' opinions, emotions, evaluations, valuations, attitudes, and emotions such as products, services, organizations, individuals, issues, activities, topics, and their attributes. Based on the results of sentiment analysis, the supply chain management decision support system can find the satisfaction level of products and services, the customer's purchase intention, product awareness and influence after marketing activities in the digital world. At the same time, for customers who prefer to purchase products, the supply chain management decision support system can also track the customer's location based on the IP address or the customer's social website status.
  • the content mining sub-module first determines whether the text is positive, neutral, or negative based on the scoring scheme.
  • the content mining sub-module scans the title and content of the text step by step; if a sentence contains keywords (for example, the name of a product or marketing campaign) and the emotional words of a predefined dictionary (different emotional words have different scores), Then the content mining sub-module will give a base score corresponding to the emotional word based on the scoring scheme.
  • the content mining sub-module may also check the prefix words of the emotional words based on the predefined dictionary to determine whether the prefix words strengthen, weaken or change the meaning of the emotional words, and give different degrees of change to the emotional words according to the prefix words. Add the component number and calculate the product of the added component number and the base score to record the score of the sentence.
  • the program of the scoring scheme is shown in Figure 2.
  • the content mining sub-module also normalizes the scores of all sentences in the text to be combined to calculate the score of the text; the content mining sub-module is also used to decrement the text related to the product or marketing activity according to the score of the text.
  • the content mining sub-module can also compare the score of the text with a preset mood threshold to obtain the emotion expressed by the text. In this embodiment, if the score of the text is greater than the preset positive emotion threshold, the emotion expressed by the text is positive. If the score of the text is greater than the preset negative emotion threshold and less than the preset positive emotion threshold, the emotion expressed by the text is neutral; if the score of the text is less than the preset negative emotion threshold, the emotion expressed by the text is negative.
  • the content mining sub-module is also used to calculate and monitor the subjectivity of text related to the product or marketing campaign.
  • the subjectivity S of the text associated with the product or marketing campaign is:
  • S is the subjectivity of the text associated with the product or marketing campaign
  • PA is the number of texts with positive emotions
  • NA is the number of texts with negative emotions
  • ZA is the number of texts with neutral emotions.
  • the present invention also defines the polarity P of the text associated with the product or marketing campaign:
  • P is the polarity of the text associated with the product or marketing campaign
  • PA is the number of texts with positive emotions
  • NA is the number of texts with negative emotions
  • the content mining sub-module is used to calculate the polarity P of the text associated with the product or marketing campaign; if the polarity of the relevant text of a product or marketing campaign is higher, the product or marketing campaign will cause the customer to have more positive emotions .
  • Customer behavior is often captured by self-hosted e-commerce based on big data technology.
  • the content mining sub-module is able to combine all user behaviors. For example, if a customer frequently searches for and views a product of a style or color, the content mining sub-module will assume that the style or color is the customer's favorite. Similarly, based on the product's click-through rate, the content mining sub-module can also determine if the customer is interested in the product. Other data, including total purchases, unprocessed items in the basket, one-day login time, landing time, etc., can also be analyzed to determine customer behavior.
  • Clustering techniques classify similar text into categories without any predefined classification.
  • the similarity of text content can be calculated, and similar content will be classified into one category.
  • the content mining sub-module can identify "hot topics," top fashion, market trends, and related topics/companies/persons.
  • the similarity between the two texts d 1 and d 2 can be calculated by the cosine similarity formula:
  • w 1k is the weight of the word k in the text d 1 ;
  • w 2k is the weight of the word k in the text d 2 ;
  • n is the total number of all words of the two texts d 1 and d 2 .
  • FIG. 3 shows the basic algorithm for agglomerative hierarchical clustering. After determining the cohesive set and word frequency, the weight of each word in each cluster can be calculated. All words for each cluster can be classified according to how their weights are decremented. Threads in cluster C i T i represents the five words the highest weight, respectively [word i1, word i2, word i3, I4 word, word i5].
  • the clustering process is different for different purposes.
  • the text being analyzed should use the text of the most recent period of time and focus on the text of the news or social media.
  • the time frame of the analyzed text should be longer (for example, one month) and the text from fashion-related news and social media needs to be analyzed.
  • Competitor information on the website can be crawled through the e-commerce website and the main competitor online store.
  • data such as product categories, product descriptions, prices, rankings, inventory levels, discounts, and the like can be obtained.
  • the supply chain management decision support system can compare these data with company data to gain competitive advantage over major competitors such as product prices, free shipping, promotions, and value-added services.
  • the supply chain management analysis engine is used to analyze the data sent by the information acquisition module from the aspect of the effectiveness and specificity of the industrial supply chain management, so as to give a supply chain management analysis report;
  • the supply chain management analysis engine consists of five main modules: the demand planning module, the production planning and sequencing module, the distribution planning module, the transportation planning module, and the enterprise or supply chain analysis module.
  • the demand planning module is used to analyze the data sent by the information acquisition module by means of statistical tools, causal elements and hierarchical analysis to predict the demand of the product.
  • the production planning and sequencing module is used to analyze the constraints of the materials and capabilities of the internal and supplier production facilities, and to prepare the production schedule of the product based on the demand of the products predicted by the demand planning module.
  • the Distribution Planning module is used to develop a distribution plan based on the product's production schedule to ensure that the product is orderable, profitable, and usable.
  • the Transportation Planning module is used to determine the best way to deliver products to customers.
  • Enterprise or supply chain analysis modules are used to display graphical models of the enterprise or supply chain to help companies strategically function in factories and distribution The heart adjusts to analyze the supply chain, pay attention to and explore problems.
  • the decision support sub-module is used to store the decision rules and solutions corresponding to the supply chain management analysis report respectively; and is also used to provide a corresponding solution to the supply chain management analysis report based on the corresponding decision rules. As shown in Figure 4, the decision support sub-module is also used to analyze the data of the supply chain management analysis report from procurement factors, production factors, time factors, market factors and inventory factors to determine product positioning.
  • the value of the purchasing factor is determined by three factors: the order form, the procurement cycle, and the cost of the purchase.
  • Orders for all raw materials used in the production of the product must be evaluated.
  • the value of the order form is determined by evaluating the following three sub-factors:
  • Order lead time The waiting time of the unspent raw material order form needs to be evaluated to determine whether the ordered raw materials are sufficient on the expected arrival date.
  • the value of the production factor is determined by three factors: production capacity, production cost, and production lead time.
  • Production capacity is determined by assessing two sub-factors (ie human resource capacity, machine capacity) to accommodate production growth.
  • the value of market factors is determined by three factors: demand forecasting, industry trends, and competitive factors.
  • Demand forecasting is determined by evaluating the following six sub-factors:
  • the promotion role the effect of the promotion (if there is a promotion);
  • customer behavior including the specific requirements and characteristics of the customer's product (including color, size And style) and the city or country where the product is relatively popular.
  • the competitive factors are determined by evaluating the following three sub-factors:
  • hot topic determined by demand forecast
  • the value of the inventory factor is determined by four sub-factors: storage capacity, storage cost, inventory level, and due date.
  • the inventory level represents the current inventory status of the company; the inventory level is determined by two sub-factors: the level of work progress and the level of raw materials.
  • the value of the time factor is determined by the product life cycle and seasonal attributes.
  • the support module also includes three sub-modules, namely the knowledge base, the textile taxonomy database and the access control sub-module, and provides support for the operation and intelligence analysis of the decision support sub-module through the three sub-modules.
  • the knowledge base is used to provide historical data for the information acquisition module and the analysis processing module for completing the analysis task;
  • the textile classification database is used to provide a textile classification method for the information acquisition module and the analysis processing module for completing the analysis task.
  • the access control sub-module is used to assign rights to the account on the login supply chain management decision support system.
  • the account number includes an administrator account and a normal account. The administrator account and the normal account have different permissions, and the range of information that can be accessed is different. Here, the permissions of the normal account are controlled by the administrator account.
  • FIG. 5 shows a flow chart of product demand forecast adjustment.
  • the information acquisition module acquires data of a big data source, the content mining sub-module mines product-related information from the data of the big data source, and the supply chain management analysis engine has information related to the product and transaction data and internal systems from the members of the supply chain. Data analysis to give a supply chain management analysis report; decision support sub-module pin Corresponding solutions are provided for the supply chain management analysis report to achieve the adjustment of product demand forecast.
  • the visualization module is used to display the analysis results and provide the following analysis reports for supply chain management decisions:
  • Product analysis reports Cost analysis reports, channel relationships and competency assessment reports, supply chain status reports, customer sentiment reports, and customer behavior reports.
  • the visualization module can display the subjectivity and polarity of the text associated with the product or marketing campaign, the text associated with the product or marketing campaign, in descending order of the text.
  • the decisions made after the analysis include the selection of sales strategies for companies such as textile companies. For example, the company will launch a new range of casual sports shoes and use advertising (such as TV commercials, magazine ads and flyers) to promote their products for more than a month. The company is willing to choose the right sales strategy for the new product; the determination of this decision needs to be done by an expert in one field.
  • companies such as textile companies. For example, the company will launch a new range of casual sports shoes and use advertising (such as TV commercials, magazine ads and flyers) to promote their products for more than a month. The company is willing to choose the right sales strategy for the new product; the determination of this decision needs to be done by an expert in one field.
  • Experts also need to check the stock status. Experts will investigate inventory factors such as raw material levels and storage capacity. Other factors reviewed by experts include product lifecycle and seasonal attributes, which is one of the notable drivers for achieving accurate product demand estimates.
  • the big data technology based supply chain management decision support system of the present invention can provide the following supply chain management services:
  • the supply chain management decision support system can make recommendations based on the product positioning analysis results for the time of reordering services, reordered content, reordered locations, and reordered quantities;
  • the supply chain management decision support system can remind the user to take corresponding actions based on the customer sentiment report when the user's extreme emotions are discovered.
  • the supply chain management decision support system can also show the user which cost is inefficient when the cost analysis report shows that the cost is not efficient, such as the raw material cost is higher than the previous contract standard; in order to find the opportunity to save costs, the user It will look for ways to save costs and increase the efficiency of cost.
  • the supply chain management decision support system can also improve the channel relationship based on the channel relationship and capability assessment report, reward or remove the channel members, and find out what channel members make suggestions.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种基于大数据技术的供应链管理决策支持系统，包括：信息获取模块，用于从大数据源提取数据，并转换其格式，再发送给分析处理模块；分析处理模块，用于从信息获取模块所发送的数据中搜索有用的信息，并对其集成和分析，从而提供分析结果；可视化模块，用于显示分析结果；以及支持模块，包括知识库、纺织品分类法数据库和访问控制子模块。该供应链管理决策支持系统能够发现用户极端情绪，并基于客户情绪报告提醒用户采取相应的行动，还能寻找可能节省成本的途经，从而提高成本使用效率。

Description

基于大数据技术的供应链管理决策支持系统 技术领域
本发明涉及信息处理领域，尤其涉及一种基于大数据技术的供应链管理决策支持系统。
背景技术
现有信息系统(如电子数据交换系统，即EDI)被用来实现信息共享，从而促进供应链有效决策的确定。
供应链经理主要依据供应链成员的交易数据和内部系统数据来针对管理、规划和操作任务做出高质量的商业或管理决策。然而，其他来源的有用数据，尤其是诸如推特、博客、新闻、邮件以及脸书的数据并没有作为依据而被分析和参与决策制定。
发明内容
本发明针对诸如推特、博客、新闻、邮件以及脸书的数据并没有作为依据而被分析和参与供应链管理决策制定的问题，提出了一种供应链管理决策支持系统。
本发明提出的技术方案如下：
本发明提出了一种供应链管理决策支持系统，包括：
信息获取模块，用于从大数据源提取数据，并转换其格式，再发送给分析处理模块；
分析处理模块，用于从信息获取模块所发送的数据中搜索有用的信息，并对其集成和分析，从而提供分析结果；
可视化模块，用于显示分析结果；以及
支持模块，包括知识库、纺织品分类法数据库和访问控制子模块；知识库 用于为信息获取模块和分析处理模块提供历史数据，以用于完成分析任务；纺织品分类法数据库用于为信息获取模块和分析处理模块提供纺织品分类法，以用于完成分析任务；访问控制子模块用于对登陆供应链管理决策支持系统上的账号进行权限分配。
本发明上述的供应链管理决策支持系统中，所述信息获取模块包括：
内容过滤器，用于滤除大数据源中的噪声数据；
文本预处理单元，用于解析由内容过滤器过滤后的数据，以得到文本；
变址单元，用于在文本中寻找并确定关键词；
命名实体识别单元，用于对文本中的词语按照预定义类别进行定位和分类；
域分类单元，用于将文本按照预定义规则划分到不同领域中；
内容评分单元，用于对文本的与纺织与时尚产业的相关性进行评分。
本发明上述的供应链管理决策支持系统中，变址单元用于通过TF-IDF算法计算多个词语在文本中的权重，并取权重最大的预设数量的词语作为关键词。
本发明上述的供应链管理决策支持系统中，所述分析处理模块包括内容挖掘子模块、供应链管理分析引擎和决策支持子模块；
内容挖掘子模块用于处理文本，并从中发现有用的信息；
供应链管理分析引擎用于对由信息获取模块所发送的数据进行分析，从而给出供应链管理分析报告；
决策支持子模块用于存储分别与供应链管理分析报告对应的决策规则和解决方案；还用于对供应链管理分析报告基于对应决策规则给出对应解决方案。
本发明上述的供应链管理决策支持系统中，内容挖掘子模块用于逐句扫描文本的标题和内容，并基于评分方案对文本句子中的情感词语给出对应的基础分数，还对情感词语的前缀词语进行检查，并按照前缀词语对情感词语的改变程度给予不同的加成分数，再计算加成分数与基础分数的乘积，以此记为该句子的分数；还用于将文本中所有句子的分数加合并进行归一化，从而计算得到 文本的分数；
内容挖掘子模块还用于将与产品或营销活动相关文本按照文本的分数递减排列；可视化模块用于显示该按照文本的分数递减排列的与产品或营销活动相关文本。
本发明上述的供应链管理决策支持系统中，内容挖掘子模块还用于将文本的分数与预设积极情绪阈值以及预设消极情绪阈值进行比较；若文本的分数大于预设积极情绪阈值时，则文本具有积极情绪；若文本的分数大于预设消极情绪阈值，并小于预设积极情绪阈值，则文本具有中立情绪；若文本的分数小于预设消极情绪阈值，则文本具有消极情绪；
内容挖掘子模块还用于计算与产品或营销活动相关文本的主观性：
Figure PCTCN2017101202-appb-000001
其中，S为与产品或营销活动相关文本的主观性；
PA为具有积极情绪的文本的数量；
NA为具有消极情绪的文本的数量；
ZA为具有中立情绪的文本的数量；
可视化模块用于显示所述主观性。
本发明上述的供应链管理决策支持系统中，内容挖掘子模块还用于计算与产品或营销活动相关文本的极性：
Figure PCTCN2017101202-appb-000002
其中，P为与产品或营销活动相关文本的极性；
PA为具有积极情绪的文本的数量；
NA为具有消极情绪的文本的数量；
可视化模块用于显示所述极性。
本发明的基于大数据技术的供应链管理决策支持系统能够提供以下供应链管理服务：1)供应链管理决策支持系统可基于产品定位分析结果给重新排序服务的时间、重新排序的内容、重新排序的位置以及重新排序的数量提出建议；2)供应链管理决策支持系统可在发现用户极端情绪时，基于客户情绪报告提醒用户采取相应的行动。3)供应链管理决策支持系统还可在成本分析报 告显示成本使用没有效率时能够为用户展示哪种成本是没有效率的，如原材料成本高于以前合同的标准；为了找寻节省成本的机会，用户会寻找可能节省成本的途经，从而提高成本使用效率。
附图说明
下面将结合附图及实施例对本发明作进一步说明，附图中：
图1示出了本发明实施例的供应链管理决策支持系统的功能模块示意图；
图2示出了本发明实施例的供应链管理决策支持系统的评分方案的程序代码图；
图3示出了凝聚层次聚类的程序代码图；
图4为图1所示的决策支持子模块所获取的数据分类图；
图5示出了产品需求预测调整的流程图。
具体实施方式
为了使本发明的技术目的、技术方案以及技术效果更为清楚，以便于本领域技术人员理解和实施本发明，下面将结合附图及具体实施例对本发明做进一步详细的说明。
如图1所示，基于大数据技术的供应链管理决策支持系统包括：信息获取模块(用于信息提取、转换和加载)、分析处理模块、支持模块以及可视化模块。
信息获取模块用于从大数据源提取数据，并转换其格式，发送给分析处理模块；大数据源可以包括其他数据库(如ERP和POS)、供应链成员以及互联网等。信息获取模块从大数据源中所获取的数据包括结构化数据和非结构化数据。在这里，结构化数据定义为可以用数据库二维逻辑表来逻辑表达实现的数据，如行数据等；而非结构化数据定义为不方便用数据库二维逻辑表来逻辑表达实现的数据，如所有格式的办公文档、文本、图片、标准通用标记语言下的子集XML、HTML、音频和视屏等。对于结构化数据，信息获取模块可以完成诸如标准化、校验、编码、合并以及拆分等转换过程，从而实现格式的转换。 对于非结构化数据，本发明主要聚焦于文本，并且，所获取的文本内容主要采用中文或英文。具体地，信息获取模块包括：
内容过滤器，用于滤除大数据源中的噪声数据。从大数据源中提取出来的数据可能会包含低质量内容，如拼写错误、无法实现的内容(如谣言)以及恶意内容，这些低质量的内容(即噪声数据)需要在被解析之前滤除掉。在本实施例中，通过检查数据源(如检查数据源是否是可信的网站等)、基于其他文本的查重检验以及使用约定俗成的知识和语言学规则可实现滤除噪声数据。
文本预处理单元，用于解析由内容过滤器过滤后的数据，以得到文本；具体地，文本预处理单元会执行清理和解析任务，如停用词的移除、分割、找出词干或者词形还原；文本预处理单元还用于将文本以一种适合分析的结构(如向量空间模型)存储下来。
变址单元，用于在文本中寻找并确定关键词；在本实施例中，变址单元通过词语在文本中的权重来寻找关键词。该权重是通过TF-IDF算法计算得到，即
Figure PCTCN2017101202-appb-000003
其中，wi,j表示词语i在文本j中的权重；
tfi,j表示词语i在文本j中的词频；
idfi表示词语i的逆文档频率；
N表示文本的总数目；
ni表示具有词语i的文本的数目。
然后，变址单元取权重最大的预设数量的词语作为关键词。
可以理解，变址单元可以将文本j表示为对应的向量
Figure PCTCN2017101202-appb-000004
有：
Figure PCTCN2017101202-appb-000005
其中，
Figure PCTCN2017101202-appb-000006
为与文本j对应的向量；
win,j表示词语in在文本j中的权重。
这样，非结构化数据就可以转换成结构化数据。
命名实体识别单元，用于对文本中的词语按照预定义类别进行定位和分 类；例如，命名实体识别单元可以将词语按照人名、组织名、地理位置名、产品名或产品类别以及情感表达进行归类，并为这些词语标识相应的释文。例如，对于“2015年太阳裙新款已在优衣库有售”这句话，在命名实体识别单元处理后，会变为“2015年[时间]太阳裙[产品类别]新款已在优衣库[公司]有售”。其中，可以看到，时间和公司会被识别出来，而词语“太阳裙”会基于纺织行业分类法以产品类别进行标示。
域分类单元，用于将文本按照预定义规则划分到不同领域中；例如，基于预定义规则，文本会划分到社交媒体(包括微博、脸书以及推特)，新闻和电子商务。
内容评分单元，用于对文本的与纺织与时尚产业的相关性进行评分。在这里，评分过程是基于预定义规则来实现的，例如，若文本中具有与纺织相关的关键词，则该文本将被评价较高分数；反之，具有在预定义规则定义的“不想要”的关键词的文本将被评价较低分数。然后，内容评分单元会将文本按照分数高低进行排列，以用于进一步的分析。
分析处理模块，用于从信息获取模块所发送的数据中搜索有用的信息，并对其集成和分析，从而提供分析结果；该分析处理模块包括三个子模块：内容挖掘子模块、供应链管理(即Supply Chain Management)分析引擎和决策支持子模块。
其中，内容挖掘子模块用于挖掘非结构化数据中有用的信息，具体来说，是用于从信息获取模块所发出的数据中确定数据内容、竞争优势和意见。具体地，内容挖掘子模块可用于处理文本，并从中发现有用的信息。
在本实施例中，基于信息获取模块所发送的数据，内容挖掘子模块通过算法可以确定客户情绪、内容群集和竞争优势。
A、客户情绪的确定
情绪分析涉及观点挖掘，并对客户针对诸如产品、服务、组织、个体、问题、活动、话题及其属性的观点、情绪、评价、估价、态度以及情感进行分析。基于情绪分析的结果，供应链管理决策支持系统可以发现产品和服务的满意水平、客户的购买意向、产品的认知度以及在数字世界的营销活动后的影响力等。 同时，对于倾向于购买产品的客户，供应链管理决策支持系统也能基于IP地址或客户的社交网站状态来追踪客户的位置。
因为由信息获取模块所发送的文本涉及产品或营销活动，内容挖掘子模块首先会根据评分方案判断文本是否是积极的、中立的或消极的。内容挖掘子模块会逐句扫描文本的标题和内容；如果某一句子包含有关键词(例如，产品或营销活动的名字)和预定义词典的情感词语(不同的情感词语具有不同的分数)，那么内容挖掘子模块会基于评分方案给出一个与情感词语对应的基础分数。内容挖掘子模块还可以基于预定义词典对情感词语的前缀词语进行检查，来判断该前缀词语是否对情感词语进行了加强、减弱或改变词意，并按照前缀词语对情感词语的改变程度给予不同的加成分数，并计算加成分数与基础分数的乘积，以此记为该句子的分数。在这里，评分方案的程序如图2所示。
内容挖掘子模块还会将文本中所有句子的分数加合并进行归一化，从而计算得到文本的分数；内容挖掘子模块还用于将与产品或营销活动相关文本按照文本的分数递减排列。优选地，内容挖掘子模块还可以将文本的分数与预设情绪阈值进行比较，从而得到文本所表达的情绪。在本实施例中，若文本的分数大于预设积极情绪阈值时，则文本所表达的情绪是积极的。若文本的分数大于预设消极情绪阈值，并小于预设积极情绪阈值，则文本所表达的情绪是中立的；若文本的分数小于预设消极情绪阈值，则文本所表达的情绪是消极的。
为了获知产品或营销活动能否引起客户的反响或讨论，内容挖掘子模块还用于计算和监控与产品或营销活动相关文本的主观性。在本实施例中，与产品或营销活动相关文本的主观性S为：
Figure PCTCN2017101202-appb-000007
其中，S为与产品或营销活动相关文本的主观性；
PA为具有积极情绪的文本的数量；
NA为具有消极情绪的文本的数量；
ZA为具有中立情绪的文本的数量。
如果由信息获取模块所发送的与某一产品或营销活动相关文本中具有较多的情绪或情感内容时，则与该产品或营销活动相关文本的主观性较高，这也 就意味着有很多客户就该产品或营销活动表达了观点或情感。
为了对产品或营销活动的情绪分析进行量化，本发明还定义了与产品或营销活动相关文本的极性P：
Figure PCTCN2017101202-appb-000008
其中，P为与产品或营销活动相关文本的极性；
PA为具有积极情绪的文本的数量；
NA为具有消极情绪的文本的数量；
内容挖掘子模块用于计算与产品或营销活动相关文本的极性P；如果某一产品或营销活动的相关文本的极性越高，则该产品或营销活动会使客户具有更多的积极情绪。
B、客户行为
客户行为通常会被自托管电商基于大数据技术捕捉到。基于大数据技术，内容挖掘子模块能够将所有用户行为合并。例如，如果一个客户经常搜寻和查看一个样式或颜色的产品，那么内容挖掘子模块将认为，该样式或颜色是该客户的最爱。同样地，基于产品的点击率，内容挖掘子模块也能判断客户是否对该产品有兴趣。包括购买总量、购物篮中的未处理项目、一天登陆时间、登陆时长等其他数据也能够被分析来确定客户行为。
C、内容群集
群集技术在没有任何预定义分类的情况下可将相似文本归为一类。文本内容的相似度可以被计算出来，相似的内容将被归为一类。因此，基于近一段时间内文本的内容群集结果，内容挖掘子模块可识别“热门话题”、顶尖时尚、市场趋势以及相关话题/公司/人物。两个文本d1和d2的相似度可通过余弦相似度公式计算得到：
Figure PCTCN2017101202-appb-000009
其中，s(d1,d2)为两个文本d1和d2的相似度；
w1k为词语k在文本d1中的权重；
w2k为词语k在文本d2中的权重；
n为两个文本d1和d2所有词语的总数目。
在供应链管理决策支持系统中，我们采用凝聚层次聚类，这是一种自下而上的群集方法。图3示出了凝聚层次聚类的基本算法。在确定了凝聚集和词频后，每个群集的每个词语的权重能够被计算出来。每个群集的所有词语可根据其权重递减方式来分类。群集Ci中的话题Ti表示权重最高的五个词语，分别为[词语i1，词语i2，词语i3，词语i4，词语i5]。
基于不同目的，群集过程也不同。如检测“热门”话题，被分析的文本应该采用最近一段时间的文本，并且要聚焦于新闻或社交媒体的文本。如要检测顶尖时尚，被分析的文本的时间范围应该要更长一些(例如，一个月)，并且，需要分析来自时尚相关新闻和社交媒体的文本。
D、竞争优势
通过电商网站和主要竞争者在线商店，网站上的竞争者信息可以被抓取。通过利用内容挖掘技术，诸如产品分类、产品描述、价格、排名、存货水平、折扣等数据可被获取。因此，供应链管理决策支持系统能够将这些数据和公司数据进行比较，从而获知相比于主要竞争者的竞争优势，如产品价格、免运费、促销活动以及增值服务等。
供应链管理分析引擎用于从工业供应链管理有效性方面和特殊性方面对由信息获取模块所发送的数据进行分析，从而给出供应链管理分析报告；
一般地，供应链管理分析引擎包括5个主要模块：需求计划模块、生产计划和排序模块、分销计划模块、运输计划模块以及企业或供应链分析模块。其中，需求计划模块用于采用统计工具、因果要素和层次分析等手段对由信息获取模块所发送的数据进行分析，从而预测产品的需求。生产计划和排序模块用于分析企业内部和供应商生产设施的物料和能力的约束，并基于需求计划模块所预测的产品的需求，编制产品的生产进度计划。分销计划模块用于基于产品的生产进度计划，制定分销计划，从而保证产品可订货、可盈利以及可使用。运输计划模块用于确定将产品送达客户的最佳途径。企业或供应链分析模块用于显示企业或供应链的图示模型，从而帮助企业从战略功能上对工厂和分销中 心进行调整，从而对供应链进行分析，注意和发掘问题。
决策支持子模块用于存储分别与供应链管理分析报告对应的决策规则和解决方案；还用于对供应链管理分析报告基于对应决策规则给出对应解决方案。如图4所示，决策支持子模块还用于从采购因素、生产因素、时间因素、市场因素以及库存因素对供应链管理分析报告的数据进行分析，从而确定产品定位。
采购因素
采购因素的数值由三个因素确定：订货单、采购周期以及采购成本。
产品生产用的所有原材料的订货单必须被评估。订货单的数值通过评估以下三个子因素确定：
1、在途：当原材料库存即将不足时，公司尚未偿付的原材料订货单需要被评估，从而确定原材料在途订货单是否充足以满足生产需要。
2、订货提前期：尚未偿付的原材料订货单的等待时间需要被评估，从而确定订购原材料在预计到达日期是否充足。
3、替代品：当原材料库存或订货单即将出现短缺时，原材料可能的替代品需要被确定。
生产因素
生产因素的数值由三个因素确定：生产能力、生产成本以及生产提前期。生产能力通过评估两个子因素(即人力资源能力、机器能力)确定，以适应生产的增长。
市场因素
市场因素的数值通过以下三个因素确定：需求预测、行业趋势和竞争因素。
需求预测是通过评估以下六个子因素来确定：
1、价格；
2、市场规模；
3、市场占有率；
4、促销作用：即促销的效果(如果有促销的话)；
5、客户行为；包括客户对产品提出的具体要求和特点(包括颜色、尺寸 和风格)以及产品相对流行的城市或国家。
6、客户情绪；主要是对产品给出积极评价的客户数量；
7、销售业绩；来自B2C平台驱动程序的销售点数据；
竞争因素是通过评估以下三个子因素来确定：
1、竞争优势(诸如产品价格、免运费、促销活动和附加业务)；
2、市场占有率；
3、市场规模。
行业趋势是通过以下两个子因素来确定：
1、顶尖时尚：用于确定新产品的可能性；
2、热门话题：由需求预测确定；
库存因素
库存因素的数值由以下四个子因素来确定：仓储能力、仓储成本、库存水平以及到期日。库存水平代表公司目前库存状态；库存水平通过以下两个子因素来确定：工作进度水平和原材料水平。
时间因素
时间因素的数值通过产品生命周期和季节性属性来确定。
支持模块也包括三个子模块，即知识库、纺织品分类法数据库和访问控制子模块，并通过该三个子模块给决策支持子模块提供操作和智力分析提供支持。其中，知识库用于为信息获取模块和分析处理模块提供历史数据，以用于完成分析任务；纺织品分类法数据库用于为信息获取模块和分析处理模块提供纺织品分类法，以用于完成分析任务；访问控制子模块用于对登陆供应链管理决策支持系统上的账号进行权限分配。所述账号包括管理员账号和普通账号，管理员账号和普通账号具有不同的权限，所能访问的信息范围是不同的。在这里，普通账号的权限受到管理员账号的控制。
如图5所示，图5示出了产品需求预测调整的流程图。信息获取模块获取大数据源的数据，内容挖掘子模块从大数据源的数据中挖掘与产品相关的信息，供应链管理分析引擎对与产品相关的信息以及来自供应链成员的交易数据和内部系统数据进行分析，从而给出供应链管理分析报告；决策支持子模块针 对供应链管理分析报告给出对应解决方案，从而实现产品需求预测的调整。可视化模块用于显示分析结果，为供应链管理决策提供以下分析报告：
产品分析报告、成本分析报告、渠道关系和能力评估报告、供应链状态报告、客户情绪报告和客户行为报告。
在上述报告中，可视化模块可以显示按照文本的分数递减排列的与产品或营销活动相关文本、与产品或营销活动相关文本的主观性以及极性。
决策实施例
在分析后做出的决策包括为公司(如纺织公司)选择销售策略。如公司会推出新的一系列休闲运动鞋，并采用广告(如电视广告、杂志广告和传单广告)来宣传他们的产品，并持续超过1个月。公司愿意为新产品选择合适的销售策略；这种决策的确定需要聘用一个领域的专家完成。
具体地，专家首先需要收集信息并检测市场因素。部分信息收集过程可采用本发明的供应链管理决策支持系统完成，包括客户意见和要求、新产品相对流行的区域、竞争状况、休闲运动鞋的顶尖时尚性和销售表现。
专家还要检测库存状况。专家会调查库存因素，如原材料水平和仓储能力。专家复核的其他因素包括产品生命周期和季节性属性，这是实现产品预估需求精确性的显著驱动器之一。
专家然后会获得如下结论：新产品需求很高(即，许多客户对新产品满意，促销活动非常成功，最后一个月的销售表现很好)，客户最喜爱的颜色和风格分别是红色和风格A，杭州是休闲运动鞋最受欢迎的城市。因此，专家会建议，公司应该增加红色和风格A的休闲运动鞋的库存水平，尤其是杭州的库存水平。
本发明的基于大数据技术的供应链管理决策支持系统能够提供以下供应链管理服务：
1)供应链管理决策支持系统可基于产品定位分析结果给重新排序服务的时间、重新排序的内容、重新排序的位置以及重新排序的数量提出建议；
2)供应链管理决策支持系统可在发现用户极端情绪时，基于客户情绪报告提醒用户采取相应的行动。
3)供应链管理决策支持系统还可在成本分析报告显示成本使用没有效率时能够为用户展示哪种成本是没有效率的，如原材料成本高于以前合同的标准；为了找寻节省成本的机会，用户会寻找可能节省成本的途经，从而提高成本使用效率。
4)供应链管理决策支持系统还可基于渠道关系和能力评估报告给渠道关系改进，渠道成员奖励或移除，寻找怎么样的渠道成员提出建议。
应当理解的是，对本领域普通技术人员来说，可以根据上述说明加以改进或变换，而所有这些改进和变换都应属于本发明所附权利要求的保护范围。

Claims (7)

  1. 一种供应链管理决策支持系统，其特征在于，包括：
    信息获取模块，用于从大数据源提取数据，并转换其格式，再发送给分析处理模块；
    分析处理模块，用于从信息获取模块所发送的数据中搜索有用的信息，并对其集成和分析，从而提供分析结果；
    可视化模块，用于显示分析结果；以及
    支持模块，包括知识库、纺织品分类法数据库和访问控制子模块；知识库用于为信息获取模块和分析处理模块提供历史数据，以用于完成分析任务；纺织品分类法数据库用于为信息获取模块和分析处理模块提供纺织品分类法，以用于完成分析任务；访问控制子模块用于对登陆供应链管理决策支持系统上的账号进行权限分配。
  2. 根据权利要求1所述的供应链管理决策支持系统，其特征在于，所述信息获取模块包括：
    内容过滤器，用于滤除大数据源中的噪声数据；
    文本预处理单元，用于解析由内容过滤器过滤后的数据，以得到文本；
    变址单元，用于在文本中寻找并确定关键词；
    命名实体识别单元，用于对文本中的词语按照预定义类别进行定位和分类；
    域分类单元，用于将文本按照预定义规则划分到不同领域中；
    内容评分单元，用于对文本的与纺织与时尚产业的相关性进行评分。
  3. 根据权利要求2所述的供应链管理决策支持系统，其特征在于，变址单元用于通过TF-IDF算法计算多个词语在文本中的权重，并取权重最大的预设数量的词语作为关键词。
  4. 根据权利要求3所述的供应链管理决策支持系统，其特征在于，所述分析处理模块包括内容挖掘子模块、供应链管理分析引擎和决策支持子模块；
    内容挖掘子模块用于处理文本，并从中发现有用的信息；
    供应链管理分析引擎用于对由信息获取模块所发送的数据进行分析，从而给出供应链管理分析报告；
    决策支持子模块用于存储分别与供应链管理分析报告对应的决策规则和解决方案；还用于对供应链管理分析报告基于对应决策规则给出对应解决方案。
  5. 根据权利要求4所述的供应链管理决策支持系统，其特征在于，内容挖掘子模块用于逐句扫描文本的标题和内容，并基于评分方案对文本句子中的情感词语给出对应的基础分数，还对情感词语的前缀词语进行检查，并按照前缀词语对情感词语的改变程度给予不同的加成分数，再计算加成分数与基础分数的乘积，以此记为该句子的分数；还用于将文本中所有句子的分数加合并进行归一化，从而计算得到文本的分数；
    内容挖掘子模块还用于将与产品或营销活动相关文本按照文本的分数递减排列；可视化模块用于显示该按照文本的分数递减排列的与产品或营销活动相关文本。
  6. 根据权利要求5所述的供应链管理决策支持系统，其特征在于，内容挖掘子模块还用于将文本的分数与预设积极情绪阈值以及预设消极情绪阈值进行比较；若文本的分数大于预设积极情绪阈值时，则文本具有积极情绪；若文本的分数大于预设消极情绪阈值，并小于预设积极情绪阈值，则文本具有中立情绪；若文本的分数小于预设消极情绪阈值，则文本具有消极情绪；
    内容挖掘子模块还用于计算与产品或营销活动相关文本的主观性：
    Figure PCTCN2017101202-appb-100001
    其中，S为与产品或营销活动相关文本的主观性；
    PA为具有积极情绪的文本的数量；
    NA为具有消极情绪的文本的数量；
    ZA为具有中立情绪的文本的数量；
    可视化模块用于显示所述主观性。
  7. 根据权利要求6所述的供应链管理决策支持系统，其特征在于，内容挖掘子模块还用于计算与产品或营销活动相关文本的极性：
    Figure PCTCN2017101202-appb-100002
    其中，P为与产品或营销活动相关文本的极性；
    PA为具有积极情绪的文本的数量；
    NA为具有消极情绪的文本的数量；
    可视化模块用于显示所述极性。
PCT/CN2017/101202 2016-10-10 2017-09-11 基于大数据技术的供应链管理决策支持系统 WO2018068603A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610886206.8 2016-10-10
CN201610886206.8A CN107918818B (zh) 2016-10-10 2016-10-10 基于大数据技术的供应链管理决策支持系统

Publications (1)

Publication Number Publication Date
WO2018068603A1 true WO2018068603A1 (zh) 2018-04-19

Family

ID=61891860

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/101202 WO2018068603A1 (zh) 2016-10-10 2017-09-11 基于大数据技术的供应链管理决策支持系统

Country Status (2)

Country Link
CN (1) CN107918818B (zh)
WO (1) WO2018068603A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636044A (zh) * 2018-12-17 2019-04-16 优必爱信息技术（北京）有限公司 一种智能网联车配件需求预测方法
CN109711639A (zh) * 2019-01-16 2019-05-03 辽宁工程技术大学 一种绿色供应链协同设计与优化系统及方法
CN110781250A (zh) * 2019-10-23 2020-02-11 陕西华筑科技有限公司 一种基于大数据的bi决策管理系统和方法
CN113792200A (zh) * 2021-09-26 2021-12-14 上汽通用五菱汽车股份有限公司 数据可视化管理系统、方法、设备及计算机可读存储介质
CN114971497A (zh) * 2022-07-06 2022-08-30 国网浙江省电力有限公司金华供电公司 基于协同互联的物资供应链全链感知的数据处理方法
CN117973805A (zh) * 2024-03-28 2024-05-03 国网浙江省电力有限公司金华供电公司 采购检测储备配送融合协同供应链管控方法及系统

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108733160A (zh) * 2018-05-10 2018-11-02 南京思达捷信息科技有限公司 一种大数据下的支持系统及其方法
CN109885011B (zh) * 2019-03-22 2020-12-08 安徽省伊贝雅纺织有限公司 一种纺织印染厂物联网管理系统
CN111026804A (zh) * 2019-12-04 2020-04-17 深圳瑞力网科技有限公司 一种基于语义的大数据分析智能服务系统
CN111369382A (zh) * 2020-02-28 2020-07-03 上海纺织建筑设计研究院有限公司 纺织智能数据处理系统
CN112115704B (zh) * 2020-09-11 2024-06-04 深圳市菲尔兹数据分析师事务所有限公司 基于电商平台的产品查询方法、装置和计算机设备
CN113657977B (zh) * 2021-10-21 2022-07-26 广州市格利网络技术有限公司 基于产业互联网的智能采购推荐方法及装置
CN114444896B (zh) * 2021-12-31 2023-03-31 浙江大学 供应链数据处理系统和方案
CN115018207B (zh) * 2022-08-04 2022-11-11 共幸科技(深圳)有限公司 一种基于上下游的供应链管理方法、系统和设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102214342A (zh) * 2010-04-09 2011-10-12 香港纺织及成衣研发中心 智能时装销售预测系统
CN105260899A (zh) * 2015-10-27 2016-01-20 清华大学深圳研究生院 一种电商主体可信度评价方法及系统
CN105677622A (zh) * 2016-03-11 2016-06-15 郑州师范学院 大数据分析报告自动生成系统
CN105741067A (zh) * 2016-01-26 2016-07-06 东华大学 一种基于大数据的纺织面料知识管理方法
WO2016119449A1 (zh) * 2015-01-26 2016-08-04 香港纺织及成衣研发中心有限公司 用于制衣生产的快速响应管理系统及方法
CN105956882A (zh) * 2016-05-09 2016-09-21 陈包容 一种获取采购需求的方法及装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8180713B1 (en) * 2007-04-13 2012-05-15 Standard & Poor's Financial Services Llc System and method for searching and identifying potential financial risks disclosed within a document
CN100520782C (zh) * 2007-11-09 2009-07-29 清华大学 一种基于词频和多元文法的新闻关键词抽取方法
CN103455562A (zh) * 2013-08-13 2013-12-18 西安建筑科技大学 一种文本倾向性分析方法及基于该方法的商品评论倾向判别器
CN105787064A (zh) * 2016-03-01 2016-07-20 广州铭诚计算机科技有限公司 一种基于大数据的挖掘平台构建方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102214342A (zh) * 2010-04-09 2011-10-12 香港纺织及成衣研发中心 智能时装销售预测系统
WO2016119449A1 (zh) * 2015-01-26 2016-08-04 香港纺织及成衣研发中心有限公司 用于制衣生产的快速响应管理系统及方法
CN105260899A (zh) * 2015-10-27 2016-01-20 清华大学深圳研究生院 一种电商主体可信度评价方法及系统
CN105741067A (zh) * 2016-01-26 2016-07-06 东华大学 一种基于大数据的纺织面料知识管理方法
CN105677622A (zh) * 2016-03-11 2016-06-15 郑州师范学院 大数据分析报告自动生成系统
CN105956882A (zh) * 2016-05-09 2016-09-21 陈包容 一种获取采购需求的方法及装置

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636044A (zh) * 2018-12-17 2019-04-16 优必爱信息技术（北京）有限公司 一种智能网联车配件需求预测方法
CN109636044B (zh) * 2018-12-17 2021-03-30 优必爱信息技术（北京）有限公司 一种智能网联车配件需求预测方法
CN109711639A (zh) * 2019-01-16 2019-05-03 辽宁工程技术大学 一种绿色供应链协同设计与优化系统及方法
CN109711639B (zh) * 2019-01-16 2022-12-20 辽宁工程技术大学 一种绿色供应链协同设计与优化系统及方法
CN110781250A (zh) * 2019-10-23 2020-02-11 陕西华筑科技有限公司 一种基于大数据的bi决策管理系统和方法
CN113792200A (zh) * 2021-09-26 2021-12-14 上汽通用五菱汽车股份有限公司 数据可视化管理系统、方法、设备及计算机可读存储介质
CN114971497A (zh) * 2022-07-06 2022-08-30 国网浙江省电力有限公司金华供电公司 基于协同互联的物资供应链全链感知的数据处理方法
CN117973805A (zh) * 2024-03-28 2024-05-03 国网浙江省电力有限公司金华供电公司 采购检测储备配送融合协同供应链管控方法及系统

Also Published As

Publication number Publication date
CN107918818A (zh) 2018-04-17
CN107918818B (zh) 2022-01-21

Similar Documents

Publication Publication Date Title
WO2018068603A1 (zh) 基于大数据技术的供应链管理决策支持系统
Zhang et al. Deciphering word-of-mouth in social media: Text-based metrics of consumer reviews
US8620718B2 (en) Industry specific brand benchmarking system based on social media strength of a brand
Liu et al. Analyzing changes in hotel customers’ expectations by trip mode
Lu et al. BizSeeker: a hybrid semantic recommendation system for personalized government‐to‐business e‐services
US10685181B2 (en) Linguistic expression of preferences in social media for prediction and recommendation
Jiwani et al. Exploring Business intelligence capabilities for supply chain: a systematic review
US9911131B1 (en) Method and system for obtaining leads based on data derived from a variety of sources
US20140358810A1 (en) Identifying candidates for job openings using a scoring function based on features in resumes and job descriptions
Li et al. Economical user-generated content (UGC) marketing for online stores based on a fine-grained joint model of the consumer purchase decision process
US20130173583A1 (en) Keyword index pruning
US20160171590A1 (en) Push-based category recommendations
Park Framework for sentiment-driven evaluation of customer satisfaction with cosmetics brands
Gupta et al. Detecting and characterizing extremist reviewer groups in online product reviews
Zhang et al. Examining differences among book reviews from various online platforms
Wu et al. Determining the factors affecting customer satisfaction using an extraction-based feature selection approach
Almagrabi et al. A survey of quality prediction of product reviews
Li et al. Restaurant recommendation model using textual information to estimate consumer preference: evidence from an online restaurant platform
Ahuja et al. Corporate blogs as tools for consumer segmentation-using cluster analysis for consumer profiling
Nie et al. A methodology for classification and validation of customer datasets
KR102404247B1 (ko) 고객 관리 시스템
JP2020201819A (ja) ビジネスマッチング支援装置、及びビジネスマッチング支援方法
Özyirmidokuz et al. Analyzing customer complaints: a web text mining application
Namvar A Novel Approach to Predict the Helpfulness of Online Reviews.
Shikalgar et al. Online Review Mining for forecasting sales

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17859470

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17859470

Country of ref document: EP

Kind code of ref document: A1