JP2006146374A - Knowledge information processor and knowledge information processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new technique for generating information expressed in a data format with high versatility as knowledge information. <P>SOLUTION: The knowledge information processor 100 comprises an input part 102 inputting table data containing a plurality of data, a location setting part 112 setting location specification information for specifying the location of each data contained in the input table data in the table data, an association part 116 associating the location of each data specified by the location specification information with the data concerned, and a knowledge information generation part 118 generating each of the associated data as one piece of knowledge information. According to this, knowledge expressed in a highly versatile table data format can be generated as knowledge information. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a knowledge information processing apparatus and a knowledge information processing method that perform inference using stored knowledge information.

  In recent years, expert systems that accumulate specialized knowledge and know-how specific to a specific field in a database, make inferences based on the accumulated specialized knowledge and know-how, and can make judgments close to experts in that field are attracting attention. It is being done. For example, a patient who has little knowledge about a disease can use the database of a doctor's accumulated knowledge about the disease to obtain the help of a doctor who is an expert.

In general, specialized knowledge and know-how prior to accumulation in a database are often expressed in a format that is difficult for a computer to understand. For example, Patent Document 1 discloses a technique for registering CAD (Computer Aided Design) data such as piping instrumentation diagrams, equipment structure diagrams, and piping diagrams that cannot be automatically registered with user input assistance.
JP-A-6-83678

  Certainly, according to Patent Document 1, information described in CAD data can be registered in a database, but the input target is limited to CAD data, and information such as knowledge and know-how expressed in other data formats. May not be applicable. Furthermore, since the use of CAD data is limited due to its nature, it may not necessarily be a highly versatile data format. Therefore, there is a demand for a new technique that can register information expressed in a highly versatile data format in a database.

  The present invention has been made in view of these problems, and an object thereof is to provide a knowledge information processing apparatus and a knowledge information processing method capable of registering information expressed in a highly versatile data format in a database as knowledge information.

  One embodiment of the present invention relates to a knowledge information processing apparatus. This device includes an input unit that inputs table data including a plurality of data, and a location setting unit that sets location specifying information for uniquely specifying the location of each data included in the input table data. And an associating unit that associates the location of each piece of data specified by the location specifying information with the data, and a knowledge information generating unit that generates each piece of associated data as one piece of knowledge information. According to this aspect, since information expressed in a highly versatile table data format can be generated as knowledge information, a knowledge information processing apparatus with high user merit can be provided.

  The location setting unit includes a key generation unit that generates a first key for specifying an item of row position in the table data and a second key for specifying an item of column position. It may be expressed two-dimensionally by the item of the row position specified by the key and the item of the column position specified by the second key.

  The input unit inputs a plurality of table data, and the location setting unit may further set identification information for specifying any one of the plurality of input table data. At least one of the first key and the second key may be formed of a plurality of keys.

  The table data is a combination of a plurality of pieces of partial table data that are significant alone, and the location specifying information may be set for each partial table data by the location setting unit. The apparatus further includes a definition information generation unit that generates definition information of the input table data, and a verification unit that compares the knowledge information with the definition information to verify whether the knowledge information is normalized. Also good.

  The associating unit selects a plurality of pieces of data in which at least the item at the row position specified by the first key is the same, and the item at the column position specified by the second key is the same among the plurality of associated data. The knowledge information generation unit includes an extraction unit for extracting, and generates a piece of knowledge information by integrating a plurality of data extracted by the extraction unit. Each piece of knowledge information may be expressed in the same format. The format may be a predicate logic type language format. The format may be an XML language format.

  Another aspect of the present invention also relates to a knowledge information processing apparatus. The apparatus includes a table data input unit for inputting table data including a plurality of data, a first key and a column position for specifying an item of a row position in the table data of each data included in the input table data. Two-dimensionally expressed by a key generation unit for generating a second key for specifying the item, a row position item specified by the first key, and a column position item specified by the second key An associating unit that associates the location of each piece of data with the data, a knowledge information generating unit that generates each piece of associated data as one piece of knowledge information, and an inference rule described in a predicate logical language format Inference rule input section, search expression input section for inputting a query search expression, inference section for performing inference based on the inference rules and knowledge information based on the query search expression, and inference by the inference section And a presentation unit that presents.

  Yet another embodiment of the present invention relates to a knowledge information processing method. The method includes inputting table data including a plurality of data, setting location specifying information for uniquely specifying the location of each data included in the input table data, and A step of associating the location of each piece of data specified by the specific information with the data, and a step of generating each piece of associated data as one piece of knowledge information.

  ADVANTAGE OF THE INVENTION According to this invention, a knowledge information processing apparatus and method with a high user merit can be provided.

Embodiment 1
FIG. 1 shows a configuration of a knowledge information processing apparatus according to the first embodiment. This configuration can be realized in hardware by a CPU (Central Processing Unit), memory, and other LSI (Large Scale Integration) in any computer, and in software, a GUI (Graphical User Interface) function and knowledge information It is realized by a program having a processing function and other functions, but here, functional blocks realized by their cooperation are depicted. Accordingly, those skilled in the art will understand that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof.

  The knowledge information processing apparatus 100 according to Embodiment 1 associates each data included in the input table data with each location of the data, and generates the associated data as knowledge information. At this time, the generated knowledge information is expressed in the same format. Knowledge information refers to information including not only “formal knowledge” that is merely data but also “implicit knowledge” that is not normally specified, such as the location of data. The knowledge information processing apparatus 100 executes inference using knowledge information stored therein and presents the inference result to the user when a user requests a problem solution by a query retrieval formula. Thereby, the knowledge information processing apparatus 100 can provide the user with a solution that satisfies the problem solving request.

  The tabular data in the first embodiment is data expressed in a tabular format such as an Excel format that is easy to visually understand when arranged two-dimensionally by rows and columns. As another example, data held in a so-called CSV (Comma Separated Value) format in which data is separated by comma symbols and data rows are separated by line feeds may be indicated. In any case, the table data may be in a data format in which the location of the data is two-dimensionally characterized by rows and columns.

  Knowledge information processing apparatus 100 according to Embodiment 1 includes an input unit 102, a presentation unit 104, a knowledge base unit 106, a definition information holding unit 108, and a control unit 110. The control unit 110 includes a location setting unit 112, an association unit 116, a knowledge information generation unit 118, an inference unit 120, a definition information generation unit 122, and a verification unit 124. The input unit 102 receives input of table data from the user via an input device (not shown) such as a keyboard, a mouse, a button, and a touch panel. Furthermore, the input unit 102 receives an input of a question retrieval formula and an inference rule from the user. In this case, the input unit 102 includes a table data input unit (not shown), a search expression input unit, and an inference rule input unit, each of which accepts input of table data, a query search expression, and an inference rule separately. Good.

  The table data does not need to be digitized, and the storage format thereof does not matter, for example, the data written on a paper document. When the table data is described on a paper document, the knowledge information processing apparatus 100 first reads the document on which the table data is described with an external scanner (not shown) as electronic drawing data. Next, an external character recognition conversion device (not shown) recognizes characters and data structures in the table data read as drawing data, and converts the table data into the CSV format. Finally, the input unit 102 accepts input of CSV format table data generated in this way. Thereby, the table data on the paper document can be input.

  The location setting unit 112 sets location specifying information for uniquely specifying the location of the plurality of pieces of data included in the table data input by the input unit 102. The location specifying information may be set according to an instruction from the user. The location specifying information includes, for example, a first key (hereinafter, simply referred to as “column key”) for specifying a row position item in the data table data and a column position item in the data table data. This refers to the second key (hereinafter simply referred to as “row key”). The row position refers to the vertical position on the table data, while the column position refers to the horizontal position on the table data.

  The key generation unit 114 in the location setting unit 112 generates a column key and a row key. The generation of the column key and the row key may be performed according to an instruction from the user. Thereby, the location of the data in the table data is two-dimensionally expressed by the item of the row position specified by the column key and the item of the column position specified by the row key. Further, the location setting unit 112 may further set identification information such as a table ID unique to the table data and a name of the table data for specifying any one of the input table data. . Thereby, even if there are a plurality of table data input to the input unit 102, the location of the data can be uniquely specified by further setting the identification information of the table data. The identification information may be set as location specifying information.

  The associating unit 116 associates the location of each piece of data specified by the location specifying information with the data. For example, if the location of the data “A” included in the table data is specified by the location specifying information as the table ID “1”, the row position “3rd row”, and the column location “4th column” The associating unit 116 associates the table ID “1”, “third row”, and “fourth column” with the data “A”.

  The knowledge information generation unit 118 generates each piece of data associated by the association unit 116 as one piece of knowledge information. The knowledge information generated by the knowledge information generation unit 118 is stored in the knowledge base unit 106. Each piece of knowledge information in the first embodiment is expressed by a “gimt” function of the same format shown below. According to the first embodiment, the knowledge information is described in the same format. Therefore, knowledge information can be easily managed by the database administrator. “Gimt” is a name of a function, but the name is not limited to this.

The definition information generation unit 122 generates definition information of table data input via the input unit 102. The definition information may be generated based on an input instruction from the user. Table data definition information is information indicating the data structure of table data, the relationship between items, or the relationship between data. The definition information generated by the definition information generation unit 122 is stored in the definition information holding unit 108. An example of definition information is shown below.
(A) Table data identification information (b) Maximum number of table data rows and maximum number of columns (c) Table data file path (d) Column number assigned column key, column key name (e) Row key Assigned line number, row key name

  The verification unit 124 compares the knowledge information generated by the knowledge information generation unit 118 with the definition information generated by the definition information generation unit 122, and verifies whether the knowledge information is normalized. The normalized state indicates a state in which data can be uniquely specified by the name of the table data and the column key and row key. Specifically, whether or not the verification unit 124 can uniquely extract knowledge information from a set of knowledge information using the name, column key, and row key of the table data included in the definition information as search keys. Accordingly, it is determined whether or not the generated knowledge information is normalized.

  The inference unit 120 performs an inference based on the knowledge information stored in the knowledge base unit 106 while referring to the inference rules when a question retrieval formula is input from the user via the input unit 102. This inference rule is described by the user and registered in the knowledge base unit 106 via the input unit 102. The inference rule may be in any description format that can use knowledge information. For example, the inference rule is expressed in a predicate logic type language format that is the same format as knowledge information. The presentation unit 104 presents the inference result from the inference unit 120 to the user via an external display (not shown).

  According to the first embodiment, the knowledge information can be uniquely specified by specifying the row position item and the column position item. That is, a single knowledge information state, a so-called normalization state, is realized for one fact, and the burden on the work of correcting or adding knowledge information can be reduced. In addition, when the row position item and the column position item are specified by the query retrieval formula, the inference unit 120 can accurately identify and use one piece of knowledge information, and therefore presents an inference result with good access efficiency and low error. it can. Thereby, the reliability of the answer to a question can be improved.

  In general, it is difficult to normalize data in a database unless someone has knowledge about database design or construction. Furthermore, even those who are involved in database design may design an inefficient structure and build a system with poor access efficiency. According to the first embodiment, knowledge information can be generated by setting at least a column key and a row key that specify the location of data. Therefore, a set of normalized knowledge information can be generated easily.

  Hereinafter, operations from inputting table data to generating knowledge information by the knowledge information processing apparatus 100 according to the first embodiment will be described with reference to FIGS. FIG. 2 shows an example of table data having a plurality of cells of 5 rows and 4 columns input to the input unit. If the table data 200 is described in the CSV format, the table data 200 converted into an easy-to-view table format such as an Excel format is shown as a diagram in consideration of ease of viewing. The table data 200 shown in FIG. 2 shows an action pattern to be taken next to a pedestrian when an event occurs at a road intersection. The name of the table data 200 is “pedestrian behavior pattern”.

  A first item group 202 composed of items from the second column to the fourth column in the first row indicates the current behavior pattern of the pedestrian such as “stop”, “walk”, and “run”. On the other hand, the second item group 204 composed of the items of the second row to the fifth row in the first column is a walk of “crossing completed” in addition to an event of a signal change of a red, blue, and blue flashing traffic light. Event related to the state transition of the person. Furthermore, the second row and the second column are the upper left corner, and the data group 206 composed of data stored in four cells in the vertical direction and three cells in the horizontal direction is as follows. The pedestrian behavior pattern to be taken next is shown according to the current pedestrian behavior pattern. It is assumed that no data is set in the cells in the first row and the first column as shown in FIG.

  In FIG. 2, for example, when an event occurs in which the current pedestrian behavior pattern is “stop” shown in the first row and second column and changes to “blue” shown in the third row and first column, It is shown that the pedestrian selects the “walking” action pattern shown in the third row and the second column as the action to be taken next. By looking at the table data 200, a human can recognize the correspondence relationship between the “current action pattern”, “the next action pattern to be taken”, and “event”. Understandable. In the first embodiment, the knowledge information processing apparatus 100 performs the following setting process so that the computer can understand the contents of the information indicated by the table data 200.

  FIG. 3 shows table data of a pedestrian behavior pattern in which location specifying information is set. The same components as those in FIG. 2 are given the same reference numerals, and description thereof will be omitted as appropriate. In addition, for ease of viewing, the cell in which the location specifying information is stored is indicated by hatching. As shown in FIG. 3, the location specifying information is stored in the cells in the 0th row, the 0th column, the 0th row, the 1st column, and the 1st row, the 0th column. The cell in which the location specifying information is stored is a cell outside the range in which the table data 200 of the pedestrian's behavior pattern is described, and thus the arrangement position of the table data 200 is not affected.

  The location specifying information may be input in a dialog form using a dialog. For example, in response to the question from the knowledge information processing apparatus 100, “What is the name of the table data?”, The user responds via the input unit 102 with the “pedestrian behavior pattern table” in the 0th row. Location specifying information may be set in the 0th column. Further, in response to the questions from the knowledge information processing apparatus 100 such as “how many columns should the column key be assigned?” And “what name should be set?”, The user may ask “1 column”, “event”. The location specifying information may be set in the 0th row and the 1st column by responding via the input unit 102. Furthermore, in this case, the set location specifying information is held in a storage unit such as a memory (not shown) in the control unit 110 and may not be displayed on the actual display.

Hereinafter, specific setting operations of location specifying information by the knowledge information processing apparatus 100 according to Embodiment 1 will be shown in order.
(1) The name “Pedestrian Behavior Pattern” of the table data 200 is set in the 0th row and the 0th column.
(2) A column key whose column key name is “event” is generated and assigned to the first column.
(3) A row key whose row key name is “current action pattern” is generated and assigned to the first row.

  That is, in (2), a column key for specifying an item of a row position of certain data included in the data group 206 is assigned to the first column, and the column key name is assigned to the second row to the fifth row of the first column. Set to "Event" shown. Similarly, in (3), a row key for specifying an item at a column position of certain data included in the data group 206 is assigned to the first row, and the row key name is assigned to the second to fourth columns of the first row. Is set to “current action pattern”. Thereby, the location of the data included in the data group 206 can be specified by the name, column key, and row key of the table data 200.

  FIG. 4 shows an example of a set of knowledge information generated by the knowledge information generation unit. For example, the knowledge information 302 in the knowledge information group 300 illustrated in FIG. 4 is obtained by converting the data “walking” in the third row and second column illustrated in FIG. 3 into knowledge. That is, the name of the table data serving as the information supply source of the knowledge information 302 is “pedestrian behavior pattern”, and when the current pedestrian behavior pattern is “stop”, the signal of the traffic light becomes “blue”. When an event occurs, the pedestrian indicates that “walk” is selected as the next action to be taken. As a result, instead of simply generating “walking” data of “formal knowledge”, knowledge information including the knowledge and know-how of “when the signal changes to blue, the pedestrian changes from a stopped state to a walking state” Can be generated.

FIG. 5 shows an example of an inference rule input by the user via the input unit. Hereinafter, the meanings indicated by the first inference rule R1 to the fourth inference rule R4 will be described. The first inference rule R1 and the second inference rule R2 are rules for performing an initialization process, while the third inference rule R3 and the fourth inference rule R4 are rules for performing an inference result output process.
(R1) When the current behavior pattern “Y” of the pedestrian is already set to some value such as “walking”, the set state is deleted and the behavior pattern of the pedestrian is set to “stop” To do.
(R2) If no value is set for the current behavior pattern “Y” of the pedestrian, the behavior pattern “Y” of the pedestrian is set to “stop”.
(R3) The event “X” and the current behavior pattern “Y” of the pedestrian are used as search keys, the knowledge information group is referenced, and there is data corresponding to the next behavior pattern “Z” that the corresponding pedestrian should take. For example, the data is output as an inference result.
(R4) There is data corresponding to the next action pattern “Z” that the corresponding pedestrian should take by referring to the knowledge information group using the event “X” and the current action pattern “Y” of the pedestrian as a search key. If not, the inference result “no rule” is output.

Hereinafter, the process until the inference unit 120 interprets and executes the first inference rule R1 to the fourth inference rule R4 and outputs the inference result will be described. The inference unit 120 may include a compiling unit (not shown), and the compiling unit may interpret and execute each inference rule.
(A) When the inference unit 120 obtains a command “init” from the user via the input unit 102, the inference unit 120 interprets and executes the first inference rule R1 and the second inference rule R2, and the pedestrian's current action pattern “ “Y” is set to the “stop” state.
(Ii) When the inference unit 120 obtains an instruction “act (blue, Y, Z)” from the user via the input unit 102, the inference unit 120 interprets and executes the third inference rule R3 and the fourth inference rule R4.
(Iii) The inference unit 120 refers to the knowledge information 302 in which the event “X” is “blue” and the pedestrian's current behavior pattern “Y” is “stop”.
(D) The inference unit 120 outputs “walk” as the next action pattern “Z” to be taken by the pedestrian as an inference result.

  FIG. 6 shows a processing flow of the knowledge information processing apparatus according to the first embodiment. The input unit 102 receives input of the table data 200 from the user (S10). The key generation unit 114 refers to the table data 200 received by the input unit 102 and generates a column key and a row key (S12). The associating unit 116 associates the location of the data specified by the column key and the row key generated by the key generating unit 114 with the data (S14). The definition information generation unit 122 generates table data definition information in accordance with an instruction from the user via the input unit 102, and stores the table data definition information in the definition information storage unit (S16).

  The knowledge information generation unit 118 generates data associated with the location as knowledge information (S18). The verification unit 124 compares the knowledge information with the definition information, and determines whether the knowledge information is normalized (S20). When it is determined by the verification unit 124 that the knowledge information is not normalized (N in S20), the presentation unit instructs the user to execute the processing after S12. When the verification unit 124 determines that the knowledge information is normalized (Y in S20), the knowledge information generation unit 118 stores the knowledge information in the knowledge base unit 106 (S22).

  The input unit 102 receives input of inference rules from the user (S24). The inference unit 120 interprets and executes the inference rules received from the user, and performs inference with reference to the knowledge information stored in the knowledge base unit 106 (S26). The presentation unit 104 presents the inference result output by the inference unit 120 to the user (S28).

  In the past, there were few companies and organizations that actively disclosed design drawings and design documents such as plant equipment and financial systems. However, in recent years, the world has accumulated as it is encouraged by the establishment of an information disclosure system. It is becoming a trend to disclose information. For this reason, companies and organizations are actively working to register such information in a database in order to publish information written in design documents and documents online. However, many of the design documents and materials incorporate tables, and in that case, the programmer may have to create a new registration program in the database using, for example, the SQL (Structured Query Language) language. is there. According to the first embodiment, a set of normalized knowledge information can be generated easily by setting at least a column key and a row key for specifying the location of the data in the table data. As a result, it is possible to greatly reduce the labor and cost for creating the program.

  Furthermore, according to the first embodiment, even a person who does not have knowledge about a database can easily register knowledge information. For example, a person in the personnel department can easily make knowledge of a table in which employment information is described. It can be registered in the base unit 106, or a person in the accounting unit can easily make a table in which financial information is described and register it in the knowledge base unit 106. As described above, by facilitating the work required for the knowledge-making work, it becomes easy to collect knowledge from experts in each field, and the range of utilization of knowledge is expanded.

  Generally, in a relational database table, data is managed in units of records, and the location of a record in a table can be uniquely specified by using a column key assigned to a field. That is, the location of the record can be specified one-dimensionally by the line position. However, the location of the data in the table data is two-dimensionally characterized by the column position and the row position. In order to uniquely identify the location of the data, it is necessary to set the column key and the row key. . In other words, the data structure of table data is different from that of a table in a relational database, so when storing table data in a table in a relational database, the two-dimensional data structure is analyzed and one-dimensional data is analyzed. A program to convert to a structure is required. According to the first embodiment, by setting at least the column key and the row key for specifying the location of data in the table data, the data included in the table data can be easily stored in the knowledge base unit. Thereby, since it is not necessary to create a conversion program, it is possible to save labor and cost for creating the conversion program.

  In general, a table in a relational database has a structure in which mere data is arranged in units of records, so that it may not be easy for a database administrator to manage the data structure. In particular, when the amount of data and the number of tables are enormous, the database management work takes more time and cost. Furthermore, when a database administrator adds or deletes additional tables in an existing database, the entire database may need to be redesigned or rebuilt. According to the first embodiment, knowledge information in a predicate logical language format that is easy for humans to understand is generated, and thus management of the knowledge base unit 106 is facilitated. In addition, knowledge information can be easily added, modified, and deleted, so that the knowledge base unit 106 can be easily expanded and maintained.

  Furthermore, the table data is used in various fields such as patient charts used by doctors, dietitian menus, train and bus timetables, and stock price tables, and can be said to be a highly versatile data format. According to the first embodiment, since the information described in the table data in the highly versatile data format can be easily converted into knowledge, the knowledge information processing apparatus 100 with high user convenience can be provided.

Embodiment 2
The number of column keys and row keys generated by the knowledge information processing apparatus 100 according to Embodiment 1 is one. However, according to the knowledge information processing apparatus 100 according to Embodiment 2, column keys and row keys are At least one of them is formed by a plurality of keys. Although details will be described later, the knowledge information processing according to the second embodiment when the row position in the data table data cannot be specified with one column key or the column position in the data table data cannot be specified with one row key. The device 100 is convenient. The configuration of knowledge information processing apparatus 100 according to Embodiment 2 is the same as the configuration of knowledge information processing apparatus 100 according to Embodiment 1. In addition, the description of the components denoted by the same reference numerals as those in Embodiment 1 is omitted as appropriate.

  The key generation unit 114 forms at least one of a column key and a row key for specifying a row position and a column position in the table data of data with a plurality of keys. In this case, the format of the knowledge information generated by the knowledge information generation unit 118 is expressed as follows. The column key group indicates a column key formed by at least one or more keys, and the row key group indicates a row key formed by at least one or more keys.

The definition information generation unit 122 generates the following definition information instead of (e) and (e) included in the definition information group shown in the first embodiment.
(F) Each column number assigned a column key group, each column key name (ki) Each row number assigned a row key group, each row key name

  Hereinafter, operations from inputting table data to generating knowledge information by the knowledge information processing apparatus 100 according to the second embodiment will be described with reference to FIGS. FIG. 7 shows an example of table data having a plurality of cells of 9 rows and 7 columns input to the input unit. The table data 210 shown in FIG. 7 is a summary of data relating to the material, type, and thickness of piping parts arranged in a plant or the like in a matrix format. The name of the table data 210 is “piping class”.

  As shown in FIG. 7, items such as “piping class” and “piping parts” are set in the first row, and further, “main body” and “ The “main part” item is set. On the other hand, the item “1P1” is displayed in the first column, the item “gate valve” or the like as the detailed item “1P1” in the second column, and the item “1 1” as the detailed item “gate valve” in the third column. / 2> "and the like are set. Furthermore, the third row and the fourth column are stored in the cells included in the range having the upper left end, the third row and the seventh column as the upper right end, the ninth row and the fourth column as the lower left end, and the ninth row and the seventh column as the lower right end. A data group 212 including data such as model names includes data such as model names.

  As shown in FIG. 7, for example, the key generation unit 114 specifies “1P1”, “gate valve”, “1 1/2 =” in order to specify the row position of the data of “scph2”, that is, “fourth row”. A column key group designating each of the three items is generated. Further, the key generation unit 114 generates a row key group for designating two items of “material” and “main body” in order to specify the column position of the data, that is, the “fourth column”.

  FIG. 8 shows tabular data of a piping class in which location specifying information is set. The same components as those in FIG. 7 are given the same reference numerals and description thereof will be omitted as appropriate. In addition, for ease of viewing, the cell in which the location specifying information is stored is indicated by hatching. As shown in FIG. 8, the location specifying information is stored in the cells in the 0th row, 0th column, the 0th row, 1st column to the 0th row, 3rd column, the 1st row, 0th column, and the 1st row, 2nd column. Yes.

Hereinafter, specific setting operations of location specifying information by the knowledge information processing apparatus 100 according to the second embodiment will be sequentially shown.
(1) The name “pipe class” of the table data 210 is set in the 0th row and the 0th column.
(2) Three column keys whose column key names are “pipe class”, “pipe part”, and “size range” are generated and assigned to the first to third columns, respectively.
(3) Three row keys whose row key names are “attribute” and “detail” are generated and assigned to the first to third rows, respectively.

  That is, in (2), three column keys included in the data group 206 for specifying the row position of certain data are assigned to the first to third columns, respectively. At this time, the column key names are set to “pipe class”, “pipe part”, and “size range” indicating items in the first column to the third column, respectively. Similarly, in (3), two row keys for specifying a column position of certain data included in the data group 206 are assigned to the first row and the second row. At this time, the row key name is set to “attribute” and “detail” indicating the item names of the first row and the second row, respectively. Thereby, the location of the data included in the data group 212 can be specified by the name, the column key group, and the row key group of the table data 210.

  FIG. 9 shows an example of a set of knowledge information generated by the knowledge information generation unit. For example, the knowledge information 312 in the knowledge information group 310 shown in FIG. 9 is obtained by converting the data in the fourth row and fourth column shown in FIG. 8 into knowledge. That is, the name of the table data serving as the information supply source of the knowledge information 312 is “piping class”, and the row position of the data is the items “1P1”, “gate valve”, “1 1/2>”, It indicates that the column position is specified by the items “material” and “main body”. As shown in FIG. 9, each data included in the data group 212 shown in FIG. 8 is generated as knowledge information in association with the location of each data.

  Table data described in a design document or a procedure manual owned by a company or organization may have a complicated data structure in which a column item is further divided into a plurality of column items. According to the second embodiment, knowledge information can be generated from table data having a complicated data structure, and the convenience of the knowledge information processing apparatus 100 is enhanced. Also in the second embodiment, the same effects as in the first embodiment can be enjoyed.

Embodiment 3
In the first embodiment and the second embodiment, a pair of column key and row key is generated for one input table data, but in the third embodiment, the table data is a significant part alone. A plurality of table data are combined, and a pair of column key and row key is generated for each partial table data. The configuration of knowledge information processing apparatus 100 according to Embodiment 3 is the same as the configuration of knowledge information processing apparatus 100 according to Embodiment 2. In addition, the description of the components having the same reference numerals as those in Embodiment 2 will be omitted as appropriate.

  The input unit 102 inputs table data obtained by combining a plurality of partial table data that are significant independently. The location setting unit 112 sets location specifying information for each partial table data. Specifically, the key generation unit 114 generates a column key and row key pair for each partial table data.

  Hereinafter, operations from inputting table data to generating knowledge information by the knowledge information processing apparatus 100 according to Embodiment 3 will be described with reference to FIGS. 10 and 11. FIG. 10 shows an example of table data having a plurality of partial table data input to the input unit. The table data 220 in FIG. 10 includes first partial table data 222, second partial table data 224, third partial table data 226, and fourth partial table data 228 as partial table data that is significant alone. The name indicated by the partial table data is the same as the name of the table data 220 formed by the partial table data, and the name is “heat exchanger”.

  For example, the key generation unit 114 generates a column key specifying the item “check” in order to specify the location of the “Nakamura” data in the first partial table data 222. For the “Nakamura” data, the location in the first partial table data 222 can be specified by the column key, so there is no need to generate a row key.

  FIG. 11 shows an example of a set of knowledge information generated by the knowledge information generation unit. The knowledge information group 320 includes a first knowledge information part group 322 to a fourth knowledge information part group 328, and each knowledge information part group includes the first part table data 222 to the fourth part table data 228 shown in FIG. It is a knowledge. As shown in FIG. 11, the knowledge information generation unit 118 clearly indicates to which partial table data the data belongs, and therefore, after the name of the table data of “heat exchanger”, the unique information corresponding to each partial table data. Knowledge information is generated by adding a unique number, for example, “1”. Thereby, normalization can be maintained.

  Taking the knowledge information 329 included in the first knowledge information subgroup 322 as an example, the information is obtained by converting the “Nakamura” data shown in FIG. 10 into knowledge. That is, the information supply source of the knowledge information 329 is the first partial table data of the table data of “heat exchanger”, and the row position of the data is specified by the item of “check”. Since the key generation unit does not generate a row key, it is represented by an underbar symbol as shown. According to the third embodiment, knowledge information can be generated from table data including a plurality of partial table data, and the versatility of the knowledge information processing apparatus is enhanced.

  Some table data described in a design document or a procedure manual owned by a company or organization may have a complicated data structure composed of a plurality of partial table data. According to the third embodiment, knowledge information can be generated from table data having a complicated data structure, and the convenience of the knowledge information processing apparatus 100 is enhanced. In the third embodiment, the same effect as in the first embodiment can be obtained.

Embodiment 4
In the first embodiment, the name of the table data and the set of column key and row key are generated as the location specifying information, but in the fourth embodiment, which column corresponds to which column is set. Specify the target data for the integration process. The configuration of knowledge information processing apparatus 100 according to Embodiment 4 is the same as the configuration of knowledge information processing apparatus 100 according to Embodiment 1. The description of the components denoted by the same reference numerals as those in Embodiment 1 is omitted as appropriate.

  The associating unit 116 extracts a plurality of pieces of data in which at least the items at the row position specified by the column key are the same and the items at the column position specified by the row key are the same from the plurality of associated data. An extraction unit (not shown) is provided. If a plurality of table data is input to the input unit 102, the associating unit 116 may extract a plurality of data having the same name of the table data as well as the column key and the row key. Although details will be described later, the knowledge information generation unit 118 integrates a plurality of data extracted by the extraction unit to generate one piece of knowledge information.

  Hereinafter, operations from inputting table data to generating knowledge information by the knowledge information processing apparatus 100 according to the fourth embodiment will be described with reference to FIGS. FIG. 12 shows an example of table data input to the input unit 102. The table data 230 shown in FIG. 12 is a table in a matrix format, meaning and usage examples for each term. The name of the table data 230 is “term dictionary”. As shown in FIG. 12, items such as “term ID” and “registration ID” are set in the first row. The column of “term ID” corresponds to the column of “term”, while “registration ID” corresponds to the column of “meaning”.

  FIG. 13 shows table data of a “term dictionary” in which location specifying information is set. The same components as those in FIG. 12 are given the same reference numerals, and description thereof will be omitted as appropriate. In addition, for ease of viewing, the cell in which the location specifying information is stored is indicated by hatching. As shown in FIG. 13, the location specifying information is stored in the cells in the 0th row and 0th column, the 0th row and 1st column to the 0th row and 4th column, and the 1st row and 0th column.

  As shown in FIG. 13, for example, the key generation unit 114 specifies “term ID”, “registration” in order to specify the row position of the data “(1) attached to the name of the mountain”, that is, “second row”. A column key group for designating two items of “ID” is generated. Further, the key generation unit 114 generates a row key for designating the item “label” in order to specify the column position of the data, that is, the “fourth column”.

  The location setting unit 112 further displays “term”, which is the name of the column key assigned to the first column, in the 0th row and the third column to indicate that the third column is associated with the first column. Set. Similarly, the location setting unit 112 further indicates the name of the column key assigned to the second column in the 0th row and the fourth column in order to indicate that the fourth column is associated with the second column. Set “Meaning”. Thereby, for example, if the “term” key is used as a search key, the data in the third column can be acquired.

  As another example, the knowledge information processing apparatus 100 designates “2” with the column key “term” and “usage example” with the row key “label”, so that “Abu Kaoru” in the fourth row and fifth column is designated. River "can be identified. On the other hand, “Kappa no river flow” in the fifth row and fifth column can be identified by the same key as “Abukuma River” in the fourth row and fifth column. That is, since the location of “Abukuma River” and “Kappa no River” is specified by the item of the same key, in this case, the knowledge information generation unit 118 includes “Abukuma River” when making the data into knowledge. The knowledge information and the knowledge information including the “Kappa River Flow” are integrated into one piece of knowledge information.

The definition information generation unit 122 further generates the following definition information in the definition information group shown in the first embodiment or the second embodiment.
(G) Location of data to be integrated, positional relationship between data to be integrated (g) Column or row number to be synchronized, column key or row key name For example, in the above example, Are “4th row, 5th column, top”, “5th row, 5th column, bottom”, and (g) is “1st column, 3rd column, term”, “2nd column, 4th” Definition information indicating “column, meaning” is generated.

  FIG. 14 shows an example of a set of knowledge information generated by the knowledge information generation unit. For example, the knowledge information 332 in the knowledge information group 330 shown in FIG. 14 is data in which the data in the fourth row, fifth column and the fifth row, fifth column shown in FIG. That is, the name of the table data serving as the information supply source of the knowledge information 332 is “term dictionary”, the row position of the data “Abukuma River, Kappa no Kawaryu” is the item “2”, while the column position is the item “Usage Example” "Indicates that it is specified.

  As shown in the figure, the location of “Abukuma River” and “Kappa River Flow” is specified by the same key item, so the knowledge information of “Abukuma River” and the knowledge information of “Kappa River Flow” are integrated, and Is generated as one piece of knowledge information data. Thus, a plurality of pieces of knowledge information do not exist for the same key item, and a normalized state can be maintained. As a result, when the inference unit 120 performs a search using a column key and a row key for a set of knowledge information, the knowledge information can be uniquely identified, and search efficiency can be improved.

  Some table data described in a design document or a procedure manual owned by a company or an organization may have a complicated data structure in which a merged cell exists. According to the fourth embodiment, knowledge information can be generated from table data having a complicated data structure, and the convenience of the knowledge information processing apparatus 100 is enhanced. Also, the fourth embodiment can enjoy the same effects as the first embodiment.

  The present invention has been described based on the embodiments. The embodiments are exemplifications, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are within the scope of the present invention. . Hereinafter, modifications will be described.

  In the first embodiment, the data group included in the table data is converted into a set of knowledge information. However, as a modified example, a table data generation unit (not shown) inside the knowledge information processing apparatus 100 refers to the definition information. Table data is created from a collection of knowledge information. Specifically, the table data generation unit first sets the table framework and items based on the information described in the definition information. Next, the table data generation unit refers to the data in the knowledge information and the location associated with the data, and stores each data in the cell specified by the associated location. Thereby, the knowledge information processing apparatus 100 can convert and present the knowledge information stored in the knowledge base unit 106 into a table format that is easier for the user to understand.

  In addition to the definition information listed in the first embodiment, the definition information generation unit 122 generates, as definition information, for example, data modes such as dotted lines and double lines, data fonts and colors, and data character sizes. May be.

  In the first embodiment, the input unit 102 inputs already completed table data. However, the user may create table data including data newly on the display and make the table data knowledgeable.

  In the first embodiment, the knowledge information is generated by converting the data included in the table data into knowledge, but as a modified example, the knowledge information generation unit 118 of the knowledge information processing apparatus 100 is data whose location is associated by the association unit 116. Are converted into knowledge information in an XML language format. Note that an XML (eXtensible Markup Language) conversion unit (not shown) inside the knowledge information generation unit 118 may perform the above-described conversion operation. The XML conversion unit converts the data included in the table data into the XML language based on the conversion rule. The XML conversion unit may convert each piece of knowledge information generated by the knowledge information generation unit 118 into an XML language format using the following conversion rule.

(1) Each data included in the table data is converted into a data format of <tag name attribute = “value” />.
(2) The attribute indicates a column key name or a row key name. If the column key or row key is composed of multiple keys, set "attribute =" value "" for each of the multiple keys. In this case, the elements of “attribute =“ value ”” are separated by a space.
(3) The value indicates an item at a row position specified by a column key or an item at a column position specified by a row key.

  For example, in the data “ISRW” in the third row and the sixth column shown in FIG. 9, “<gimt piping class =“ 1P1 ”piping component =“ gate valve ”size range =“ 1 1/2> ”attribute =” The type “data =“ ISRW ”/>” is generated by the XML conversion unit. According to this modification, not only knowledge information described in a predicate logic language format but also knowledge information described in a highly versatile XML language format can be generated.

In the fourth embodiment, data having the same column key and row key is integrated to generate one piece of knowledge information. However, as a modification, the knowledge information generation unit 118 uses the item of the row position specified by the column key. Or you may produce | generate the knowledge information of the format which abbreviate | omitted a part of column position item specified with a row key. For example, taking FIG. 13 as an example, the knowledge information generation unit 118 may generate the following knowledge information in addition to the knowledge information group 330 shown in FIG. That is, some items of the row position specified by the column key are omitted.
gimt ('Term Dictionary', ['3'], ['Usage Examples'], 'The Caspian Sea, which was a sea of fire, a stormy sea').

  In this case, the inference unit 120 designates “term dictionary” as the table data name, “3” as the column key item, and “usage example” as the row key item, so that “the Caspian Sea, the sea of fire is around” The data of “Arashi no Umi” can be collected together. Thereby, a knowledge information processing apparatus with high user convenience can be provided.

1 is a diagram illustrating a configuration of a knowledge information processing apparatus according to Embodiment 1. FIG. It is a figure which shows an example of the table data which has the some cell input into the input part which concerns on Embodiment 1. FIG. It is a figure which shows the table data of the pedestrian's action pattern in which the location specific information which concerns on Embodiment 1 was set. 6 is a diagram illustrating an example of a set of knowledge information generated by a knowledge information generation unit according to Embodiment 1. FIG. It is a figure which shows an example of the inference rule input via the input part by the user. 4 shows a processing flow of the knowledge information processing apparatus according to the first embodiment. It is a figure which shows an example of the table data which has the some cell input into the input part which concerns on Embodiment 2. FIG. It is a figure which shows the table data of the piping class in which the location specific information which concerns on Embodiment 2 was set. It is a figure which shows an example of the collection of the knowledge information produced | generated by the knowledge information production | generation part which concerns on Embodiment 2. FIG. It is a figure which shows an example of the table data which has the some partial table data input into the input part which concerns on Embodiment 3. FIG. FIG. 10 is a diagram illustrating an example of a collection of knowledge information generated by a knowledge information generation unit according to Embodiment 3. It is a figure which shows an example of the table data input into the input part which concerns on Embodiment 4. It is a figure which shows the table data of the "term dictionary" in which the location specific information which concerns on Embodiment 4 was set. FIG. 20 is a diagram illustrating an example of a collection of knowledge information generated by a knowledge information generation unit according to Embodiment 4.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 100 Knowledge information processing apparatus, 102 Input part, 104 Presentation part, 112 Location setting part, 114 Key generation part, 116 Association part, 118 Knowledge information generation part, 120 Inference part, 122 Definition information generation part, 124 Verification part, 200, 210, 220, 230 Table data, 222, 224, 226, 228 Partial table data, 302, 312, 329, 332 Knowledge information.

Claims (12)

An input unit for inputting table data including a plurality of data;
A location setting unit for setting location specifying information for uniquely specifying the location in the table data of each data included in the input table data;
An associating unit that associates the location of each piece of data specified by the location specifying information with the data;
A knowledge information generating unit that generates each piece of associated data as one piece of knowledge information;
A knowledge information processing apparatus comprising: The location setting unit includes a key generation unit that generates a first key for specifying an item of a row position in the table data and a second key for specifying an item of a column position,
The location of the data is two-dimensionally expressed by an item of a row position specified by the first key and an item of a column position specified by the second key. Knowledge information processing equipment.   The knowledge information processing apparatus according to claim 2, wherein at least one of the first key and the second key is formed of a plurality of keys.   The input unit inputs a plurality of table data, and the location setting unit further sets identification information for specifying any one of the input plurality of table data. The knowledge information processing apparatus according to any one of claims 1 to 3.   The table data is a combination of a plurality of pieces of partial table data that are significant independently, and the location specifying information is set for each partial table data by the location setting unit. 4. The knowledge information processing apparatus according to any one of 4 above. A definition information generator for generating definition information of the input table data;
A verification unit that compares the knowledge information with the definition information and verifies whether the knowledge information is normalized;
The knowledge information processing apparatus according to claim 1, further comprising: In the associating unit, at least the row position item specified by the first key is the same among the plurality of associated data, and the column position item specified by the second key is the same. It has an extraction unit that extracts multiple pieces of certain data,
The knowledge information processing apparatus according to claim 2, wherein the knowledge information generation unit generates a piece of knowledge information by integrating a plurality of data extracted by the extraction unit.   The knowledge information processing apparatus according to claim 1, wherein each of the knowledge information is expressed in the same format.   The knowledge information processing apparatus according to claim 8, wherein the format is a predicate logic type language format.   The knowledge information processing apparatus according to claim 8, wherein the format is an XML language format. A table data input section for inputting table data including a plurality of data;
A key generation unit that generates a first key for specifying an item of a row position in the table data of each data included in the input table data and a second key for specifying an item of a column position;
An associating unit for associating the location of each piece of data expressed two-dimensionally with the item of the row position specified by the first key and the item of the column position specified by the second key with the data;
A knowledge information generating unit that generates each piece of associated data as one piece of knowledge information;
An inference rule input part for inputting an inference rule described in a predicate logic type language format;
A search expression input section for inputting a question search expression;
An inference unit that performs inference based on the inference rule and the knowledge information based on the query retrieval formula;
A presentation unit for presenting an inference result by the inference unit;
A knowledge information processing apparatus comprising: Inputting table data including a plurality of data;
Setting location specifying information for uniquely specifying the location in the table data of each data included in the input table data;
Associating the location of each piece of data specified by the location specifying information with the data;
Generating each piece of associated data as a piece of knowledge information;
A knowledge information processing method comprising:

Images