JP2006235952A - Business model about information analysis and utilization - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new and excellent business model about a method for increasing the efficiency of a company's reception work and making effective use of available information. <P>SOLUTION: The business model stores all voice data about communication with consumers; converts the stored voice data into text data using a voice recognition technique; stores the text data in a sharable state; deals with consumer inquiries by referring to and extracting the data stored in a sharable state; and analyzes the text data so as to obtain analysis results useful for business, such as customers' attributes and tendencies. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a business model of window business in a company. In particular, the present invention relates to a business model and system for efficiently dealing with inquiries and complaints from consumers and identifying improvement points.

  In general corporate window operations, operators directly handle inquiries and complaints from consumers over the phone. At this time, a computer database may be used to improve the efficiency of business. In other words, customer information and inquiries are stored in a database and can be accessed and confirmed by multiple persons (operators).

  Further, with the development of IT technology in recent years, there are cases in which inquiries and complaints from consumers are received by e-mail via the Internet. Since inquiries by e-mail are basically text data, it is relatively easy to improve business efficiency by IT. For example, Japanese Patent Laid-Open No. 2003-141152 discloses a method and system for efficiently using information received from a user.

JP2003-141152

An object (problem) of the present invention is to provide a new and excellent business model relating to the efficiency of a company's window service and a method for effectively utilizing the obtained information.

  In order to solve the above-mentioned problems, in the business model according to the present invention, all the speech data related to the interaction with the consumer is accumulated, the accumulated speech data is converted into text data by speech recognition technology, and the text Store data in a sharable state, respond to inquiries from consumers by referring to and extracting the data stored in the sharable state, analyze the text data, and analyze customer attributes and trends Get useful analysis results for your business.

Preferably, the analysis includes the following steps.
(1) Importing the text data (2) Decomposing the imported text data into necessary words (3) Totaling the words obtained by the decomposition (4) Visually expressing the total result Do

  Embodiments of the present invention are described using the following examples.

  The present invention relates to a technique in which a company establishes a long-term relationship with a customer by applying an information system. It is realized by integrating and managing all interactions with individual customers, including complaints and opinions, as well as purchase usage history based on a detailed customer database. By responding closely to needs, it is possible to increase customer convenience and satisfaction, increase the maintenance rate, and maximize long-term profits.

  In the system that achieves the business model of the present invention, the three systems, that is, a recording system, a knowledge system, and a mining system are all operated in conjunction. When responding to consumers, use the knowledge system to give the best answer. Also, record all calls at the same time. The stored FAQ data and the recorded voice are converted into text, and the mining operation is performed by the mining system.

  According to the business model of the present invention, all conversations with consumers are recorded, and the information is converted into text so that the top of the company can grasp the current situation without alteration. In addition, the use of the knowledge system accumulates the response status and inquiry information as needed, so it is always possible to reply based on the latest information. Since the system itself leads to the answer to be answered, it is possible to provide a uniform service. Furthermore, by utilizing the mining system, it is possible to provide new services, find improvement points, and make proposals for improving CRM (Customer Relationship Management).

  The knowledge system according to the present invention is based on a management technique for improving performance by sharing and effectively utilizing knowledge and information of an individual as a whole. The knowledge information in this case refers to a wide range including not only data but also work know-how. It is possible to improve work and provide a place for innovation, such as the development of individual abilities, the productivity of the entire organization, and the speed of decision making.

  The recording system according to the present invention uses voice recognition software equipped with artificial intelligence in addition to a recording device for recording telephone call voice. In addition, the recognized voice is converted to text and linked with the knowledge system to improve the quality of the contact center, solve problems, and support various contacts.

  The mining system according to the present invention analyzes a large amount of unformatted text (natural sentences). Unlike simple document retrieval that retrieves knowledge information useful for business, the trend and correlation of the contents described in text information is judged and analyzed from various perspectives, so that it can be grasped only by using an existing database. Discover the attributes, trends, and true needs of customers that were difficult to use, and use them in planning sales strategies.

Hereinafter, the present invention will be described in more detail with reference to the drawings.
FIG. 1 is an explanatory diagram showing an outline of a business model according to the present invention. In the present invention, all voice data related to the exchange between the consumer 12 and the call center (window business department) 14 is stored in the management center (database) 18, and the stored voice data is converted into text data by voice recognition technology. , Save it in a sharable state. The call center 14 responds to inquiries from the consumer 12 by referring to / extracting data stored in the management center 18 in a sharable state. The management center 18 analyzes the accumulated text data and obtains analysis results useful for the business such as customer attributes and trends. The obtained analysis result is provided to the company manager 16 in the form of “report of recorded contents”, “indication of improvement points”, and the like. The corporate manager 16 does not necessarily have to be a manager and may be a mere person in charge.

  When the business model of the present invention is viewed in hardware, a server is installed in the management center 18, and a plurality of client terminals are installed in the call center 14. The server and the client terminal can be connected via a LAN, WAN, or public communication line.

  FIG. 2 is a block diagram showing the flow of data processing in the business model according to the present invention. When a consumer places an order for a product or the like at a call center, the consumer's data is stored as customer data. When there is an inquiry or complaint from the consumer to the call center, all calls are recorded and temporarily stored as call data. The stored call data is converted into text format or CSV format by voice recognition. The converted character data (txt, csv) is accumulated as shared knowledge. In a call center that receives a consumer inquiry or complaint, the customer data and inquiry / answer data are referred to, and the inquiry history of the corresponding consumer is searched and confirmed. If similar inquiry data / answer data already exists, the customer is referred to and the customer is answered.

  In FIG. 2, customer data, call data, and inquiry / answer data can be stored on the hard disk of the same server. Further, the voice recognition process and the data analysis (mining) can be performed on the same server where each data is stored or on another computer.

  What is stored as accumulated data can include paper documents and electronic mail exchanges in addition to the above-described audio data. Data accumulated in the database is analyzed (mined) by a method described later to obtain a desired analysis result. The analysis result is provided to the company manager 16 shown in FIG. 1 together with the recording data.

  FIG. 3 is an explanatory diagram showing a mining process in the business model according to the present invention. First, a CSV format file (or TXT format) to be analyzed is imported, and an atypical character string item to be mined is decomposed into words called concepts. As a method for decomposing into concepts, two methods of main term extraction (FIG. 4) and concept extraction (FIG. 5) can be adopted. The generated concept group is aggregated as an analysis file according to the application, and is converted into various graphs, which can be referenced (knowledge discovery) from various viewpoints. Here, import refers to “preparation work for analysis” that takes in a text file (.txt or .csv) to be analyzed when performing text mining (text analysis work).

  In FIG. 3, “generation of analysis file” includes a step of collecting the disassembled concepts and creating a concept list. Mainly narrow down concept-containing documents and aggregate attribute information of concept-containing documents. As described above, there are two methods for disassembling the analysis text: “main word extraction” and “concept extraction”. In addition, the two concept list files are compared, and the concepts that have not appeared in the past are analyzed. Washing out (new generation) can also be performed.

  “Concept co-occurrence extraction” is to graph the relationship between the concepts in the concept list generated by concept extraction using a network diagram, an overall bird's-eye view, a co-occurrence comparison diagram, and the like. Specifically, the relationship between concepts created by concept extraction is confirmed using four types of graphs. Before performing concept co-occurrence extraction, the analyst must understand the "distance between concepts".

“Distance between concepts” means, for example, “Tomorrow, we will watch the game of Daiei Hawks at Fukuoka Dome.” "
Five concepts are extracted in this order. In this case, the distance between the concepts of "Fukuoka Dome" and "Daiei Hawks" is "1". The distance between the concepts of “Fukuoka Dome” and “watching” is “3” because there are two concepts “Daiei Hawks” and “Game” between them. Based on the distance between concepts, a function that can refer to a relationship diagram between concepts is concept co-occurrence extraction. Proper nouns such as Fukuoka Dome and Daiei Hawks need to be registered in the user dictionary.

  “Clustering” refers to classifying all text-imported documents into an arbitrary number of groups. Based on the result of concept extraction, similar documents are systematically divided. Importance is given to all the concepts, so a more organized document classification is possible. There are three reference methods for clustering: 1) cluster file reference, 2) cluster overhead view display, and 3) cluster hierarchy display.

  “Category classification” refers to classification into an arbitrary category designated by the user using the concept extraction result.

  Hereinafter, concept extraction in the present invention will be described. Concept extraction is the most important function in the mining process. Breaks down text strings into concepts (words) and generates "concept files" that are the basis of various analysis tasks.

  FIG. 4 is an explanatory diagram showing the main word extraction process. The main word extraction is an operation to decompose a sentence character string to be analyzed into a concept while referring to a common main term dictionary and a user dictionary (main term tab) of the system. In addition, Katakana Kuri, Kanji Kuri, and Alphanumeric Kuri can be considered as concepts and disassembled. By referring to the deleted word tab and synonym tab of the user dictionary, editing is performed based on the user dictionary for unnecessary concepts, unusual synonyms, and the like.

  Is the XX character string a word for concept extraction? The computer performs the judgment criteria. A dictionary file is used at that time. There are two types of dictionary files (default dictionary and user dictionary), and "concept extraction" can be performed by combining these dictionaries. Basic Japanese is recorded in the default dictionary, and for example, 10,000 to 20,000 words can be recorded for main word extraction. In the user dictionary, each user can freely register an arbitrary character string as a word.

  FIG. 5 is an explanatory diagram showing the concept word extraction process. Concept extraction refers to the default dictionary (common teacup dictionary) and user dictionary (teacup word tab) of the system, and is decomposed into concepts, and the concept words (nouns + adjectives, nouns + verbs, etc.) are converted into user dictionaries (concepts). Edit based on (extraction rules). The edited concept word refers to the deleted concept word tab and the synonym concept word tab of the user dictionary, and edits unnecessary concept concepts and synonym concept words based on the user dictionary.

  As described above, according to the business model of the present invention, all conversations with consumers (users, customers) are recorded, and the information is converted into text, so that the top of the company is not falsified. It becomes possible to grasp. In addition, the use of the knowledge system accumulates the response status and inquiry information as needed, so it is always possible to reply based on the latest information. Since the system itself leads to the answer to be answered, it is possible to provide a uniform service. Furthermore, by utilizing the mining system, it is possible to provide new services, find improvement points, and make proposals for improving CRM (Customer Relationship Management).

  In the mining system according to the present invention, unlike simple document retrieval for extracting knowledge information useful for business, by analyzing and analyzing trends and correlations of contents described in text information from various viewpoints, This will bring about an epoch-making effect of discovering customer attributes, trends, and true needs that were difficult to grasp using existing databases alone, and can be used for planning sales strategies.

FIG. 1 is an explanatory diagram showing an outline of a business model according to the present invention. FIG. 2 is a block diagram showing the flow of data processing in the business model according to the present invention. FIG. 3 is an explanatory diagram showing a mining process in the business model according to the present invention. FIG. 4 is an explanatory diagram showing a main word extraction process in the mining process in the business model according to the present invention. FIG. 5 is an explanatory diagram showing a concept word extraction process in the mining process in the business model according to the present invention.

Explanation of symbols

10 system 12 consumer 14 call center 16 company manager 18 management center

Claims (3)

Accumulate all audio data related to consumer interaction,
The accumulated voice data is converted into text data by voice recognition technology,
Save the text data in a sharable state,
Responding to inquiries from consumers by referring to and extracting data stored in a sharable state,
A business model characterized by analyzing the text data and obtaining analysis results useful for business such as customer attributes and trends. The business model according to claim 1, wherein the analysis includes the following steps.
(1) Importing the text data (2) Decomposing the imported text data into necessary words (3) Totaling the words obtained by the decomposition (4) Visually expressing the total result Do The step (2) includes a step of deleting unnecessary words,
The business model according to claim 2, wherein the step (3) includes a step of calculating importance of each decomposed word.

Images