JP6751188B1 - Information processing apparatus, information processing method, and information processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processing device, an information processing method and an information processing program for finding similar documents more efficiently. SOLUTION: An information processing apparatus 100 determines a point as an integration result of a vector representing the meaning of a word included in each of a plurality of documents stored in a document database and a meaning of a word included in a newly acquired document. The difference between the first calculation unit 101, which calculates the distance from the point as the integration result of the represented vector as the first similarity, the characters included in each of the plurality of documents, and the characters included in the newly acquired document. A second calculation unit 102 that calculates as a distance and calculates the closeness of the distance as a second similarity, and a plurality of documents that are similar to the newly acquired document based on the first similarity and the second similarity. A selection unit 103 for selecting from a document is provided. [Selection diagram] Fig. 1

Description

The present invention relates to an information processing device, an information processing method, and an information processing program.

In the above technical field, in Patent Document 1, a word included in a document is assigned a degree of importance, a feature of the document is expressed by a multidimensional vector having the degree of importance of the word as an element, and it is more similar to the angle of the vector between the documents. A technique for determining the degree is disclosed.

JP, 2005-219799, A

However, with the technique described in the above document, it was not possible to efficiently find a similar document.

An object of the present invention is to provide a technique that solves the above problems.

In order to achieve the above object, an information processing apparatus according to the present invention,
A receiving unit for receiving from the user the document data as a search target including the word that will be described in the option field of the specification of the machine tool,
A calculation unit that calculates the first document vector by vectorizing and integrating the meanings of a plurality of words included in the document data;
The option column data indicating a document freely described in the option column of the past specifications and the second document vector as a result of vectorizing and integrating the meanings of the words included in the option column data are associated with each other. A specification database to be stored,
A first calculator that calculates a cosine distance between the first document vector and the second document vector as a first similarity;
A character string included in the option field data, compared with the character string included in the document data, a second calculation unit for calculating a distance representing a difference in string as a second degree of similarity,
A selection unit for selecting, from the specification database, a similar specification including the option field data similar to the document data , based on the first similarity and the second similarity.
A transmission unit that transmits the similar specifications to the user,
Equipped with.
In order to achieve the above object, another information processing apparatus according to the present invention is
An acquisition unit that acquires specification data including the type of machine, schedule, and option fields that can be freely input,
(a) A first vector as an integration result of vectors representing the meanings of words included in each option field of a plurality of specification data stored in the document database, and included in the specification data acquired by the acquisition unit. The cosine distance between the second vector as the integration result of the vector representing the meaning of the word included in the first option described in the option column described above is calculated as the first similarity of the first option, and (b) the above Cosine distance between the first vector and the third vector as the integration result of the vector representing the meaning of the word included in the second option described in the option column included in the specification data acquired by the acquisition unit A first calculation unit that calculates as the first similarity of the second option,
(c) The difference between the character string included in each option column of the plurality of specification data stored in the document database and the character string included in the first option is calculated as a distance, and the closeness of the distances is calculated. Is calculated as the second similarity of the first option, and (d) a character string included in each option column of a plurality of specification data stored in the document database, and a character string included in the second option. A second calculation unit that calculates the difference as a distance and calculates the closeness of the distance as the second similarity of the second option;
(e) selecting a document similar to the first option from a plurality of specifications stored in the document database based on the first similarity and the second similarity of the first option, and (f) A selection unit for selecting a document similar to the second option from a plurality of specifications stored in the document database based on the first similarity and the second similarity of the second option;
Equipped with.

In order to achieve the above object, an information processing method according to the present invention is
A receiving step in which the receiving unit receives document data as a search target including a word that will be described in the option field of the specification of the machine tool from the user,
A document vector calculation step in which a document vector calculation unit calculates a first document vector by vectorizing and integrating the meanings of a plurality of words included in the document data;
The option column data indicating a document freely described in the option column of the past specifications and the second document vector as a result of vectorizing and integrating the meanings of the words included in the option column data are associated with each other. A first calculation step in which a first calculation unit calculates a cosine distance between the first document vector and the second document vector as a first similarity using a specification database that stores
A second calculation step in which a character string included in the option field data is compared with a character string included in the document data, and a distance representing the difference between the character strings is calculated by the second calculator as the second similarity.
A selection step in which a selection unit selects a similar specification including the option field data similar to the document data from the specification database based on the first similarity and the second similarity;
A transmission step in which the transmission unit transmits the similar specifications to the user,
including.

In order to achieve the above object, an information processing program according to the present invention,
The points as the result of adding up the vectors representing the meanings of the words contained in each of the multiple documents stored in the document database and the points as the result of adding up the vectors representing the meanings of the words contained in the newly acquired document. A first calculation step of calculating the distance as the first similarity,
A second calculation step of calculating a difference between a character included in each of the plurality of documents and a character included in the newly acquired document as a distance, and calculating a closeness of the distance as a second similarity degree;
A selection step of selecting a document similar to the newly acquired document from the plurality of documents based on the first similarity and the second similarity;
Causes the computer to execute.

According to the present invention, a similar document can be found more efficiently.

It is a figure explaining the internal configuration of the information processing apparatus which concerns on 1st Embodiment of this invention. It is a figure explaining the outline of operation|movement of the information processing apparatus which concerns on 2nd Embodiment of this invention. It is a figure explaining an example of the similarity search algorithm of the information processing apparatus which concerns on 2nd Embodiment of this invention. It is a block diagram which shows the structure of the information processing apparatus which concerns on 2nd Embodiment of this invention. It is a figure explaining an example of the content of the reply mail returned by the information processor concerning a 2nd embodiment of the present invention. It is a figure which shows an example of the vectorization table which the information processing apparatus which concerns on 2nd Embodiment of this invention has. It is a figure which shows an example of the vector conversion table which the information processing apparatus which concerns on 2nd Embodiment of this invention has. It is a block diagram which shows the hardware constitutions of the information processing apparatus which concerns on 2nd Embodiment of this invention. It is a flow chart explaining the processing procedure of the information processor concerning a 2nd embodiment of the present invention.

Hereinafter, modes for carrying out the present invention will be described in detail by way of example with reference to the drawings. However, the configurations, numerical values, processing flows, functional elements, and the like described in the following embodiments are merely examples, and modifications and changes thereof are free, and the technical scope of the present invention is described below. It is not meant to be limiting.

[First Embodiment]
An information processing apparatus according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram for explaining the internal configuration of the information processing apparatus according to this embodiment.

The information processing device 100 includes a calculation unit 101, a calculation unit 102, and a selection unit 103. The calculation unit 101 represents a point 162 as an integration result of vectors representing the meanings of the words included in each of the plurality of documents 161 stored in the document database 160, and the meaning of the words included in the newly acquired document 150. The distance to the point 152 as the vector integration result is calculated as the similarity 111. The calculation unit 102 calculates the difference between the character 163 included in each of the plurality of documents 161 and the character 153 included in the newly acquired document 150 as the distance, and calculates the closeness of the distance as the similarity 121. .. The selection unit 103 selects the document 170 similar to the newly acquired document 150 from the plurality of documents 141 based on the similarity 111 and the similarity 121.

According to this embodiment, a similar document can be found more efficiently.

[Second Embodiment]
Next, an information processing apparatus according to the second embodiment of the present invention will be described with reference to FIGS. 2A to 6. FIG. 2A is a diagram for explaining the outline of the operation of the information processing device 200 according to the present embodiment. For example, when the machine tool seller 210 receives an estimate request for a machine tool or a design request for a machine tool from a person who wants to purchase the machine tool, a document 250 (design document or specification sheet ) is issued based on such a request. It is created and transmitted to the information processing device 200. The information processing apparatus 200 searches the database 260 for a specification 261 similar to the received document 250, and presents the specification 261 of the machine tool of a similar case in the past to the machine tool seller 210.

By referring to the specification 261 of similar cases in the past, the machine tool seller 210 can confirm what kind of parts are attached to the machine tool when making a quotation or actually designing a machine tool by receiving an order and a price. You can immediately see how much was. Therefore, the machine tool seller 210 can easily create and design an estimate, and can significantly reduce the time for creating an estimate, the time for designing, and the time required for ordering parts.

The machine tool specifications 250 and 261 include fields for entering the machine tool type, schedule (business negotiation start date, meeting date, order date), options, and the like. Here, the option field corresponds to a so-called free input field, and is a field for entering information for designating various specifications of the machine tool according to the individual circumstances of each customer. That is, in the option column, the customer's request is described in a free format. For example, one machine tool seller 210 writes the names of parts, the number of parts, etc. in a bulleted form, and another machine tool seller 210 writes the customer's request in text. As described above, the option field has a high degree of freedom, and depending on the machine tool seller 210, the option field may be described by using an abbreviation, an abbreviation, a symbol, or the like, and there is no fixed format for the description.

When creating a quotation for a machine tool or a design, it is important to describe the option column. That is, for example, even if the machine tool models are the same, the estimation contents and the design contents are significantly different depending on what option is set. Therefore, the machine tool seller 210 searches for past similar cases based on the contents described in the option column.

However, when the machine tool seller 210 considers a keyword for a search and performs a search using the keyword, the machine tool seller 210 wants if the keyword considered by the machine tool seller 210 is not appropriate. Such search results will not be returned.

Therefore, the machine tool seller 210, to create a sentence that would have been described in the options field of the past of specifications, terms, words, symbols, numbers outgoing mail 2 5 0 was current as, the information processing apparatus 200 Send to. The information processing apparatus 200 which has received the outgoing mail 2 5 0, based like in the text that is described in the outgoing mail 2 5 0, retrieves the specifications with a description of the option field similar from the database 260, and extracts. Then, the information processing apparatus 200 returns the extracted specifications as a reply mail 230 to the machine tool seller 210. The specifications of the machine tool are stored and managed in the database 260. The database 260 may be internal or external to the information processing device 200.

At the time of searching the database 260, the information processing apparatus 200 calculates the degree of similarity with the description in the option column desired by the machine tool seller 210 for all the specifications stored in the database 260, and calculates the degree of similarity. Extract high specifications. In the calculation of the similarity, the information processing apparatus 200 calculates the similarity using two indices of the cosine distance and the Levenshtein distance, as shown in FIG. 3C. The calculation method using the cosine distance is a method for calculating a high value for the similarity between words or words having similar meanings, and is a calculation method based on the meaning of words.

The calculation method using the Levenshtein distance is a method of calculating the degree of similarity based on how similar the character strings themselves are, and is a character-based calculation method. For example, when the difference between the character strings is 3 characters or less, a high similarity value is calculated, and when the difference is 4 characters or more, a low similarity value is calculated. The information processing apparatus 200 searches for past similar cases by combining these two methods of calculating the degree of similarity. The information processing apparatus 200 calculates, for example, the similarity based on the cosine distance with a perfect score of 50 points, the similarity based on the Levenshtein distance with a perfect score of 50 points, and the total of the two (a perfect score of 100 points) to calculate the overall similarity.

The information processing apparatus 200 compares the calculated similarity value with a predetermined threshold value, and determines a specification having a similarity value higher than the predetermined threshold value as a similar case. The information processing apparatus 200 attaches the specifications determined to be similar cases to the reply mail 230 and sends the reply to the machine tool seller 210. Thereby, for example, regarding the description of the option column of a certain model of the machine tool, it can be seen that there was a past specification attached to the reply mail 230, that is, a similar case in the past.

FIG. 3A is a block diagram showing the configuration of the information processing apparatus according to this embodiment. FIG. 3B is a diagram illustrating an example of the content of a reply mail returned by the information processing device according to the present embodiment. The information processing device 200 includes a calculation unit 301, a calculation unit 302, a selection unit 303, and a transmission/reception unit 304. First, attention is focused on a single specification 241 of the plurality of specifications 241 stored in the database 260. The calculation unit 301 obtains vectors representing the meanings of all the words included in the specification 241, integrates the obtained vectors, and obtains the resulting points.

Next, the calculation unit 301 performs the above calculation for all the remaining specifications stored in the database 260, and obtains a point as the integration result of the vectors of the respective specifications.

Subsequently, the computing unit 301 obtains a word vector representing the meaning of a word contained in the newly acquired document 250, a sum of word vectors determined, the point indicating the document vector as result of integration word vectors of its Find the coordinates of.

Then, the calculation unit 301 calculates the cosine distance between the point 262 as the document vector of the specification 261 and the point 252 as the document vector of the document 250 as the similarity 311. The calculated similarity 311 is temporarily stored in the calculation unit 301.

The cosine distance similarity 311 is called a cosine similarity , and is a similarity calculation method used when comparing documents in a vector space model. The similarity between two document vectors is the same. It is an index that indicates whether or not you are facing. If the cosine distance is close to 1, it means that they are similar, and if it is close to 0, it means that they are not similar.

The calculation unit 302 calculates the difference between the character string 263 included in each of the plurality of specifications 261 stored in the database 260 and the character string 253 included in the newly acquired document 250 as a distance. The calculating unit 302 calculates the closeness of the calculated distances as the similarity 321.

Here, the distance calculated by the calculation unit 302 is called a Levenshtein distance. The Levenshtein distance is a type of distance indicating how different two character strings are.

The selection unit 303 adds up the similarity 311 calculated by the calculation unit 301 and the similarity 321 calculated by the calculation unit 302 to generate a similarity 331. The selection unit 303 selects a specification similar to the document 250 as the similar specification 270 from the specification 261 stored in the database 260 based on the generated similarity 331.

The similarity 331 generated by adding the similarity 311 and the similarity 321 is a combination of the similarity based on the meaning and the similarity based on the character. By combining these two similarities in a well-balanced manner. , A similar specification 270 similar to the document 250 can be extracted.

The transmitting/receiving unit 304 attaches the selected similar specification 270 to a reply mail and sends the reply to the machine tool seller 210. Specifically, the returned email has the content shown in FIG. 3B. Note that the transmission/reception unit 304 may return the address of the link destination to the similar specification 270 to the user as a reply mail instead of attaching the selected similar specification 270 as a reply.

Note that the selected similar specification 270 may receive feedback from the machine tool seller 210 as to whether or not the machine tool seller 210 desires it so that the quality of the selection result can be learned.

Although the specification of the machine tool has been described as an example here, the newly acquired document 250 received by the information processing device 200 is not limited to the specification, and, for example, a part purchase order or a product It may be an inquiry document from the user in the case of a failure.

FIG. 4A is a diagram showing an example of a vectorization table included in the information processing apparatus according to the present embodiment. The specification ID (Identifier) 411 is an identifier for identifying the specification, and is uniquely assigned to each specification stored in the database 260. The database 260 stores the created specifications each time a new specification is created. The option column 412 is a document that freely expresses the customer's request, and describes what kind of parts are added, what the price is, and the like. The document vector 413 is the sum of the vectors of the words included in the specification (document). In this way, if the document vector in the option field of each specification is calculated in advance, the specifications having similarities can be quickly and surely specified by obtaining the cosine distance (cosine similarity) between the document vectors. be able to. In addition to these, the vectorization table 401 may store, for example, the machine tool model, schedule, NC device type, and the like. Here, the schedule includes, for example, a business negotiation start date, a business negotiation meeting date, an ordering date, and the like. The vectorization table 401 may be sorted by model.

FIG. 4B is a diagram showing an example of the vector conversion table included in the information processing apparatus according to the present embodiment. The vector conversion table 402 stores a vector 422 in association with a word 421. The word 421 is a word included in the specification of the machine tool. The word 421 includes abbreviations, foreign languages, technical terms, etc. in addition to general terms. The vector 422 is each vector of the word 421, and shows what kind of vector it will be when the word 421 is vectorized. That is, the vector conversion table 402 represents the correspondence between words and vectors. Since the information processing device 200 calculates the cosine distance for each word in advance, the cosine distance in the option field of the specification can be easily calculated by referring to the vector conversion table 402.

FIG. 5 is a block diagram showing the hardware configuration of the information processing apparatus according to this embodiment. A CPU (Central Processing Unit) 510 is a processor for arithmetic control, and executes a program to realize the functional configuration unit of the information processing apparatus 200 in FIG. The CPU 510 has a plurality of processors and may execute different programs, modules, tasks, threads, etc. in parallel. A ROM (Read Only Memory) 520 stores fixed data such as initial data and programs and other programs. The network interface 530 also communicates with other devices and the like via the network. The number of CPUs 510 is not limited to one, and may be a plurality of CPUs or may include a GPU (Graphics Processing Unit) for image processing. Further, it is desirable that the network interface 530 has a CPU independent of the CPU 510 and writes or reads transmission/reception data to/from an area of a RAM (Random Access Memory) 540. Further, it is desirable to provide a DMAC (Direct Memory Access Controller) for transferring data between the RAM 540 and the storage 550 (not shown). Further, the CPU 510 recognizes that the data is received or transferred to the RAM 540 and processes the data. Further, the CPU 510 prepares the processing result in the RAM 540, and leaves the subsequent transmission or transfer to the network interface 530 or the DMAC.

The RAM 540 is a random access memory used by the CPU 510 as a work area for temporary storage. The RAM 540 has an area reserved for storing data necessary for realizing the present embodiment. The acquired document 541 is a document transmitted by the machine tool seller 210 to the information processing apparatus 200 for searching for similar specifications. The cosine distance 542 is a distance indicating how similar the acquired document 250 and the specifications 261 stored in the database 260 are. The Levenshtein distance 543 is a distance indicating how similar the characters included in the acquired document 250 and the characters of the specifications 261 stored in the database 260 are on a character basis. The similarity 544 is a ratio determined based on the cosine distance 542 and the Levenshtein distance 543 and the similarity between the acquired document 250 and the specifications 261 stored in the database 260. The similar specification 545 is a document similar to the acquired document 250 selected based on the similarity 544.

The transmission/reception data 546 is data transmitted/received via the network interface 530. The RAM 540 also has an application execution area 547 for executing various application modules.

The storage 550 stores a database, various parameters, and the following data or programs necessary for implementing the present embodiment. The storage 550 stores the vectorization table 401 and the conversion table 402. The vectorization table 401 is a table for storing the specification ID 411 and the document vector 413 shown in FIG. 4A in association with each other. The conversion table 402 is a table that stores the words 421 and the vectors 422 shown in FIG. 4B in association with each other.

The storage 550 further stores a calculation module 551, a calculation module 552, a selection module 553, and a transmission/reception module 554. The calculation module 551 is a module that calculates the cosine distance. The calculation module 552 is a module that calculates the Levenshtein distance. The selection module 553 is a module that selects the similar specification 270 similar to the newly acquired document 250 based on the cosine distance and the Levenshtein distance. The transmission/reception module 554 is a module that receives the document 250 by e-mail and returns the similar specification 270 by e-mail. These modules 551 to 554 are read by the CPU 510 into the application execution area 547 of the RAM 540 and executed. The control program 555 is a program for controlling the entire information processing apparatus 200.

The input/output interface 560 interfaces the input/output data with the input/output device. A display unit 561 and an operation unit 562 are connected to the input/output interface 560. A storage medium 564 may be further connected to the input/output interface 560. Further, a speaker 563 that is a voice output unit, a microphone (not shown) that is a voice input unit, or a GPS position determination unit may be connected. Note that the RAM 540 and the storage 550 shown in FIG. 5 do not show programs or data relating to general-purpose functions of the information processing apparatus 200 or other feasible functions.

FIG. 6 is a flowchart for explaining the processing procedure of the information processing apparatus 200 according to this embodiment. This flowchart is executed by the CPU 510 of FIG. 5 using the RAM 540 to realize the functional configuration unit of the information processing apparatus 200 of FIG.

In step S601, the information processing apparatus 200 receives, from the user, a document in which a keyword or the like for searching for past specifications that the user wants to know is written. In step S603, the information processing apparatus 200 calculates the cosine distance between the document received from the user and the specifications stored in the database 260. In step S605, the information processing apparatus 200 calculates the Levenshtein distance between the document received from the user and the specifications stored in the database 260. In step S607, the information processing apparatus 200 calculates the similarity of the storage specifications by taking the sum of the calculated cosine distance and the calculated Levenshtein distance. In step S609, the information processing apparatus 200 determines whether or not the calculation of the similarity has been completed for all the specifications stored in the database 260. When the calculation of the similarity has not been completed for all the storage specifications (NO in step S609), the information processing apparatus 200 repeats the processing from step S603. When the calculation of the similarity has been completed for all the storage specifications (YES in step S609), the information processing apparatus 200 proceeds to step S611. In step S611, the information processing apparatus 200 attaches a specification having a degree of similarity equal to or higher than a predetermined threshold to the e-mail as a similar specification and transmits it to the user.

According to this embodiment, since the similarity is determined by combining the cosine distance and the Levenshtein distance, it is possible to quickly and surely find a past specification that is similar to the document the user is looking for.

[Other Embodiments]
Although the present invention has been described with reference to the exemplary embodiments, the present invention is not limited to the above exemplary embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the technical scope of the present invention. Further, a system or apparatus in which different features included in each embodiment are combined in any way is also included in the technical scope of the present invention.

Further, the present invention may be applied to a system including a plurality of devices or may be applied to a single device. Furthermore, the present invention is applicable to a case where an information processing program that realizes the functions of the embodiments is supplied to a system or apparatus and executed by a built-in processor. Therefore, in order to realize the functions of the present invention on a computer, both a program installed in the computer, a medium storing the program, a WWW (World Wide Web) server for downloading the program, and a processor for executing the program are provided. It is included in the technical scope of the invention. In particular, at least a non-transitory computer readable medium storing a program that causes a computer to execute the processing steps included in the above-described embodiment is included in the technical scope of the present invention.

Claims (3)

An acquisition unit that acquires specification data including the type of machine, schedule, and option fields that can be freely input,
(a) A first vector as an integration result of vectors representing the meanings of words included in each option field of a plurality of specification data stored in the document database, and included in the specification data acquired by the acquisition unit. The cosine distance between the second vector as the integration result of the vector representing the meaning of the word included in the first option described in the option column described above is calculated as the first similarity of the first option, and (b) the above Cosine distance between the first vector and the third vector as the integration result of the vector representing the meaning of the word included in the second option described in the option column included in the specification data acquired by the acquisition unit A first calculation unit that calculates as the first similarity of the second option,
(c) The difference between the character string included in each option column of the plurality of specification data stored in the document database and the character string included in the first option is calculated as a distance, and the closeness of the distances is calculated. Is calculated as the second similarity of the first option, and (d) a character string included in each option column of a plurality of specification data stored in the document database, and a character string included in the second option. A second calculation unit that calculates the difference as a distance and calculates the closeness of the distance as the second similarity of the second option;
(e) selecting a document similar to the first option from a plurality of specifications stored in the document database based on the first similarity and the second similarity of the first option, and (f) A selection unit for selecting a document similar to the second option from a plurality of specifications stored in the document database based on the first similarity and the second similarity of the second option;
Information processing device equipped with. A receiving unit for receiving from the user the document data as a search target including the word that will be described in the option field of the specification of the machine tool,
A calculation unit that calculates the first document vector by vectorizing and integrating the meanings of a plurality of words included in the document data;
The option column data indicating a document freely described in the option column of the past specifications and the second document vector as a result of vectorizing and integrating the meanings of the words included in the option column data are associated with each other. A specification database to be stored,
A first calculator that calculates a cosine distance between the first document vector and the second document vector as a first similarity;
A character string included in the option field data, compared with the character string included in the document data, a second calculation unit for calculating a distance representing a difference in string as a second degree of similarity,
A selection unit for selecting, from the specification database, a similar specification including the option field data similar to the document data , based on the first similarity and the second similarity.
A transmission unit that transmits the similar specifications to the user,
Information processing device equipped with. The information processing apparatus according to claim 1, wherein the distance indicating the difference between the character strings is a Levenshtein distance.

Images