[go: up one dir, main page]

WO2024252512A1 - Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement - Google Patents

Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement Download PDF

Info

Publication number
WO2024252512A1
WO2024252512A1 PCT/JP2023/020951 JP2023020951W WO2024252512A1 WO 2024252512 A1 WO2024252512 A1 WO 2024252512A1 JP 2023020951 W JP2023020951 W JP 2023020951W WO 2024252512 A1 WO2024252512 A1 WO 2024252512A1
Authority
WO
WIPO (PCT)
Prior art keywords
sentence
sentences
structuring
unit
classification
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/JP2023/020951
Other languages
English (en)
Japanese (ja)
Inventor
琢磨 佐藤
了 藤井
文浩 谷口
拓也 高関
嘉之 衛藤
大輔 松田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp filed Critical NEC Corp
Priority to PCT/JP2023/020951 priority Critical patent/WO2024252512A1/fr
Publication of WO2024252512A1 publication Critical patent/WO2024252512A1/fr
Anticipated expiration legal-status Critical
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/205—Parsing
    • G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars

Definitions

  • Patent Document 1 describes how predicates are extracted from source documents written in natural language, and phrases related to those terms are extracted to generate relationship information that indicates the relationships between them.
  • Patent Document 1 The technology in Patent Document 1 is based on the premise that a sentence contains a predicate, as well as phrases such as a subject and object that are related to that term. However, in sentences written in natural language, some elements such as the subject or object may be omitted. And with the technology in Patent Document 1, it is difficult to obtain appropriate relationship information, i.e., structuring results, for sentences that are missing some elements.
  • One aspect of the present invention has been made in consideration of the above problems, and one example of its objective is to provide an information processing device or the like that is capable of obtaining valid structuring results even for sentences that lack at least one of a subject and an object.
  • An information processing device includes a classification means for classifying sentences according to their contents, an estimation means for estimating at least one of the subject and object of the sentence by applying an estimation method according to the result of the classification, and a structuring means for structuring the sentence based on the result of the estimation.
  • a structuring method includes at least one processor classifying sentences according to their contents, applying a predetermined estimation method according to the results of the classification to estimate at least one of the subject and object of the sentences, and structuring the sentences based on the results of the estimation.
  • a recording medium is a computer-readable recording medium that records a structuring program that causes a computer to function as a classification means for classifying sentences according to their contents, an estimation means for estimating at least one of the subject and object of the sentences by applying a predetermined estimation method according to the results of the classification, and a structuring means for structuring the sentences based on the results of the estimation.
  • FIG. 1 is a block diagram showing a configuration of an information processing device according to a first exemplary embodiment of the present invention
  • 1 is a flow chart showing the flow of a structuring method according to an exemplary embodiment 1 of the present invention.
  • FIG. 11 is a block diagram showing a configuration of an information processing device according to an exemplary embodiment 2 of the present invention. 11 is a diagram showing an example in which a sentence is structurized by an information processing device according to an exemplary embodiment 2 of the present invention, and the result is displayed as a graph.
  • FIG. FIG. 11 is a flowchart showing a flow of processing executed by an information processing device according to a second exemplary embodiment of the present invention.
  • FIG. 11A and 11B are diagrams showing an example of element extraction by an element string extraction unit and an example of displaying the extraction results.
  • 11 is a diagram for explaining processing related to an element classification unit and an element extraction unit.
  • FIG. 11 is a flow diagram showing the flow of a process for identifying a correspondence relationship with a document.
  • FIG. 13 is a diagram showing an example display of the results of identifying the correspondence between structured sentences and research materials showing the detection results of a specific vehicle by a vehicle detection system.
  • FIG. 13 is a diagram showing an example display of the results of identifying the correspondence between elements of a structured sentence and places shown in research materials.
  • FIG. 13 is a diagram showing an example display of the results of identifying the correspondence between structured sentences and call histories shown in the research materials.
  • FIG. 13 is a diagram showing an example display of the results of identifying the correspondence between structured sentences and deposit/withdrawal histories shown in the research materials.
  • FIG. 1 is a diagram showing an example of a computer that executes instructions of a program, which is software that realizes the functions of each device according to each exemplary embodiment of the present invention.
  • Example embodiment 1 DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
  • This exemplary embodiment is a basic form of the exemplary embodiments described below.
  • Fig. 1 is a block diagram showing the configuration of the information processing device 1. As shown in the figure, the information processing device 1 includes a classification unit 11, an estimation unit 12, and a structuring unit 13.
  • the classification unit 11 classifies sentences according to their contents. For example, as will be described with reference to FIG. 3, the classification unit 11 may classify sentences according to their contents into commands, questions, greetings, sentences expressing emotions such as gratitude or apologies, and so on.
  • the estimation unit 12 applies an estimation method according to the result of classification by the classification unit 11 to estimate at least one of the subject and object of the sentence.
  • the estimation unit 12 may perform at least one of the following processes for a sentence classified as an imperative sentence: a process of estimating the sender of the sentence as the subject of the sentence, and a process of estimating the receiver of the sentence as the object of the sentence.
  • the estimation unit 12 may estimate at least one of the subject and object of a sentence classified as a sentence other than an imperative sentence, based on the sentences before and after the sentence.
  • the structuring unit 13 structures the sentence based on the result of the estimation by the estimation unit 12. For example, as will be described with reference to FIG. 3, the structuring unit 13 may use preset sentence analysis rules or analysis models to extract a subject and an object and words indicating the relationship between them from the sentence to be structured. Also, for example, the structuring unit 13 may structure a sentence using a technology called OpenIE (Open Information Extraction).
  • OpenIE Open Information Extraction
  • the information processing device 1 includes a classification unit 11 that classifies sentences according to their contents, an estimation unit 12 that estimates at least one of the subject and object of the sentence by applying an estimation method according to the result of classification by the classification unit 11, and a structuring unit 13 that structures the sentence based on the result of estimation by the estimation unit 12. Therefore, the information processing device 1 according to this exemplary embodiment has the effect of being able to obtain a valid structuring result even for a sentence that is missing at least one of the subject and object.
  • the functions of the information processing device 1 described above can also be realized by a program.
  • the structuring program according to the present exemplary embodiment is configured to cause a computer to function as a classification means for classifying sentences according to their contents, an estimation means for estimating at least one of the subject and object of the sentence by applying a predetermined estimation method according to the result of the classification, and a structuring means for structuring the sentence based on the result of the estimation.
  • the recording medium according to the present exemplary embodiment is a computer-readable recording medium, and records the structuring program. Therefore, the structuring program according to the present exemplary embodiment or the recording medium according to the present exemplary embodiment has the effect of making it possible to obtain a valid structuring result even for a sentence lacking at least one of the subject and the object.
  • Fig. 2 is a flow diagram showing the flow of the structuring method. Note that the execution subject of each step in this structuring method may be a processor provided in the information processing device 1, a processor provided in another device, or a processor provided in a different device.
  • At least one processor classifies the sentences according to their content.
  • At least one processor applies a predetermined estimation method according to the result of the classification in S11 to estimate at least one of the subject and object of the sentence.
  • At least one processor structures the sentence based on the results of the estimation in S12.
  • the structuring method according to this exemplary embodiment includes at least one processor classifying sentences according to their contents, applying a predetermined estimation method according to the results of the classification to estimate at least one of the subject and object of the sentence, and structuring the sentence based on the result of the estimation. Therefore, the structuring method according to this exemplary embodiment has the effect of making it possible to obtain a valid structuring result even for a sentence that is missing at least one of the subject and object.
  • Fig. 3 is a block diagram showing the configuration of the information processing device 2.
  • the information processing device 2 is a device having a function of structuring sentences. Note that the information processing device 2 may be a device whose main function is to structure sentences, or may be a general-purpose device having other functions as well.
  • the information processing device 2 includes a control unit 20 that controls each unit of the information processing device 2, and a storage unit 21 that stores various data used by the information processing device 2.
  • the information processing device 2 also includes a communication unit 22 that enables the information processing device 2 to communicate with other devices, an input unit 23 that receives various data input to the information processing device 2, and an output unit 24 that enables the information processing device 2 to output various data.
  • the control unit 20 of the information processing device 2 has a classification unit 201, an estimation unit 202, a structuring unit 203, a graph generation unit 204, an information presentation unit 205, an element sequence extraction unit 206, an element classification unit 207, an element extraction unit 208, and a relevance identification unit 209.
  • the memory unit 21 also stores a classification model 211, target data 212, and materials 213.
  • the element sequence extraction unit 206, element classification unit 207, element extraction unit 208, relevance identification unit 209, and document 213 will be described later with reference to Figures 6 to 12.
  • the functions of each block from the memory unit 21 to the output unit 24 may be realized by a device built into the information processing device 2, or may be realized by a device external to the information processing device 2 that is attached to the information processing device 2.
  • the information processing device 2 may have a function of associating each sentence included in the target data 212 with a label indicating its classification to create training data. Furthermore, the information processing device 2 may have a function of re-training the classification model 211 using new training data.
  • the estimation unit 202 may estimate that the subject is "Person A” and the object is "Person B.”
  • the structuring unit 203 may apply the estimation result as it is to generate a structured result of (subject: Person A, relationship: know, object: Person B).
  • the structuring unit 203 may also add the estimated subject "Person A” before the sentence "I know,” and add the estimated object "Person B” after the sentence to generate the sentence "Person A, know, Person B.”
  • the structuring unit 203 may then perform structuring on the sentence "Person A, know, Person B.”
  • the structuring unit 203 may also add the estimated subject and object before the sentence "I know.”
  • the structuring unit 203 may add particles according to the subject and object. This makes it possible to generate a natural sentence such as "Person A knows Person B.”
  • structuring a sentence means breaking down the sentence into its components and clarifying the relationships between each of the elements.
  • structuring unit 203 can also be described as an estimation means for estimating the relationships between each of the elements that make up a sentence, or an information generation means for generating information indicating the relationships between the elements contained in a sentence.
  • the structuring method is not particularly limited.
  • the structuring unit 203 may use preset sentence analysis rules or analysis models to extract a subject and an object and words indicating the relationship between them from a sentence to be structured.
  • the word indicating the relationship between the subject and the object may be, for example, a predicate.
  • the structuring unit 203 can structure a sentence using OpenIE. By using OpenIE, it is possible to extract a subject and an object and words indicating the relationship between them from a sentence.
  • the graph generation unit 204 generates a graph showing the results of structuring by the structuring unit 203.
  • the graph generation unit 204 may generate a graph in which each element constituting a sentence is represented by a node, and the relationships between the elements are represented by edges connecting the nodes.
  • the estimation unit 202 estimates at least one of the subject and the object of a sentence by applying an estimation method according to the result of classification by the classification unit 201. It is only necessary to determine in advance which estimation method the estimation unit 202 applies depending on which classification result is obtained.
  • the estimation unit 202 may perform at least one of the following processes: a process of estimating the sender of the sentence as the subject of the sentence, and a process of estimating the receiver of the sentence as the object of the sentence. This makes it possible to obtain a valid estimation result and a valid structuring result. Furthermore, with this configuration, a valid estimation result can be obtained even if there are no sentences before or after the target sentence, and the missing elements of the target sentence cannot be inferred from the sentences before and after. Note that directive sentences, threatening sentences, etc. are also included in the category of the above-mentioned "imperative sentences”.
  • the estimation unit 202 may estimate at least one of the subject and object based on the sentences before and after the sentence.
  • the method of estimating the subject and object based on the sentences before and after the sentence is not particularly limited.
  • the estimation unit 202 may estimate the omitted subject and object using an estimation model that estimates at least one of the omitted subject and object.
  • the estimation unit 202 may extract the subject and object from the sentence before or after the sentence in which at least one of the subject and object is omitted, and use the extracted subject and object as the estimation result. Note that when extracting the subject and object in a sentence, a technique such as NER may also be applied.
  • Example of processing 4 is a diagram showing an example in which a sentence is structured by the information processing device 2 and the results are displayed as a graph.
  • the sentence to be structured is a sentence sent from sender A to receiver B, and the text of the sentence is "Please submit it immediately.”
  • This sentence does not include a subject, and does not state who it is to be submitted to, so it is difficult to structure this sentence as it is.
  • the structuring unit 203 structures the above sentence based on the above estimation result.
  • the structuring unit 203 may set the subject and object estimated by the estimation unit 202 as the subject and object of the above sentence.
  • the structuring unit 203 then extracts the character string "Please submit", which is an element indicating the relationship between the subject and object supplemented by the estimation unit 202, from the above sentence. This results in a structured result of (subject: sender A, relationship: "Please submit", object: receiver B).
  • Fig. 5 is a flow diagram showing an example of the process executed by the information processing device 2.
  • the classification unit 201 acquires the target data 212.
  • the target data 212 may include, for example, multiple sentences that are related to each other. For example, messages sent and received by a specific person may be the target data 212. In this case, multiple sentences related to the person become the target data 212.
  • the classification unit 201 reads one sentence from the target data 212 acquired in S21. Then, in S23, the classification unit 201 classifies the sentence read in S22 according to its content. For example, as described above, the classification unit 201 may perform classification using the classification model 211.
  • the estimation unit 202 applies a predetermined estimation method according to the result of the classification in S23 to the sentence read in S22, and estimates at least one of the subject and object.
  • the structuring unit 203 structures the sentence read out in S22 based on the estimation result in S24.
  • the structuring unit 203 may generate multiple structuring results from one sentence. For example, if the target sentence is "A Corporation is located in Tokyo, the capital of Japan," the structuring unit 203 may generate two structuring results: (subject: A Corporation, relationship: is, object: Tokyo) and (subject: capital of Japan, relationship: is, object: Tokyo).
  • the classification unit 201 determines whether structuring has been completed for all sentences contained in the target data 212 acquired in S21. If the determination in S26 is NO, the process returns to S22, and a new sentence is read from the target data 212. On the other hand, if the determination in S26 is YES, the process proceeds to S27.
  • the graph generation unit 204 generates a graph showing the structuring result of S25. Then, in S28, the information presentation unit 205 displays the graph generated in S27 on the display device. This ends the processing in FIG. 5.
  • the graph generation unit 204 may link the multiple graphs via a node that is common to the multiple graphs. For example, suppose that the graph generation unit 204 generates a graph in which a node called "subject 1" is connected to a node called "object 1" via an edge, and a graph in which a node called “subject 1” is connected to a node called “object 2" via an edge. In this case, the graph generation unit 204 may link these graphs via the node called "subject 1" that is common to these graphs. As a result, a graph is generated in which a node called “object 1" is connected to a node called "object 2" via an edge to the node called "subject 1".
  • the process of FIG. 5 described above includes a structuring method according to this exemplary embodiment. That is, the structuring method according to this exemplary embodiment includes classifying sentences according to their contents (S23), inferring at least one of the subject and object of the sentence by applying a predetermined inference method according to the result of the classification in S23 (S24), and structuring the sentence based on the inference result in S24 (S25).
  • S23 contents
  • S24 predetermined inference method according to the result of the classification in S23
  • S25 structuring the sentence based on the inference result in S24
  • the element string extraction unit 206 extracts a series of elements that are associated based on the structuring result and the classification result by the classification unit 201, from among the elements that constitute the multiple sentences that have been structured.
  • the series of elements is represented as a series of nodes connected by edges in the graph generated by the graph generation unit 204. Therefore, it can be said that the element string extraction unit 206 extracts a series of nodes connected by edges.
  • the element string extraction unit 206 may extract one edge corresponding to a sentence that the classification unit 201 has classified into a certain category (e.g., a command sentence), and extract each element connected by the extracted edge.
  • the element string extraction unit 206 may repeat the process of extracting other elements connected to each extracted element by an edge of the certain category until no new elements are extracted. In this way, a series of elements that are associated based on the structuring result and the classification result are extracted.
  • the information processing device 2 equipped with the element string extraction unit 206 has the effect of being able to extract a series of related elements using the classification results of each sentence.
  • the classification results of a sentence can also be considered as the classification results of the edge corresponding to that sentence. Therefore, it can be said that the element string extraction unit 206 extracts a series of related elements using the classification results of the edge.
  • the element string extraction unit 206 may accept specification of conditions for the extraction target.
  • the element string extraction unit 206 may accept specification of a category of sentences classified by the classification unit 201.
  • the element string extraction unit 206 extracts a series of elements connected by edges from among the sentence elements of the specified category.
  • FIG. 6 shows an example of element extraction by the element string extraction unit 206 and an example of the display of the extraction results. More specifically, FIG. 6 shows graph G1 generated by the graph generation unit 204 and graph G1' showing the element extraction results by the element string extraction unit 206. A user can generate and display a graph like graph G1 by, for example, inputting messages exchanged by persons A to D into the information processing device 2 as target data 212.
  • the classification unit 201 may classify command sentences into subcategories such as instructions, threats, and orders. This makes it possible to specify the extraction target on a subcategory basis. For example, it is possible to extract elements corresponding to sentences classified as orders.
  • the designation may be received via the input unit 23 or the communication unit 22.
  • the element string extraction unit 206 having received the above specification, extracts a series of elements connected by edges from among the nodes and edges corresponding to the sentences classified as imperative sentences by the classification unit 201, from the structuring results shown in graph G1. Specifically, in the example of FIG. 6, the element string extraction unit 206 extracts each of the elements "Person A" to "Person D" and the edges connecting them (edges corresponding to sentences classified as imperative sentences).
  • Graph G1' shows the results of this extraction reflected on graph G1.
  • the nodes of "Person A” to "Person D” and the edges of the command statements connecting them are highlighted with thicker lines than the other nodes and edges. In this way, the nodes corresponding to the extracted elements and the edges connecting these nodes are highlighted on the graph, allowing the user to recognize the results of the extraction.
  • graph G1' shown in FIG. 6 it can be easily inferred that Person A is at the top of the chain of command, Persons C and D are at the bottom of the chain of command, and Person B is the intermediary between them. Note that the manner in which the results of the extraction are presented is not limited to this example, as long as it allows the user to recognize the results of the extraction.
  • the element classification unit 207 classifies each element constituting a sentence into a plurality of categories. These plurality of categories may include a person category.
  • the classification method is not particularly limited.
  • the element classification unit 207 may classify each element constituting a sentence by applying a method such as NER (Name Entity Recognition).
  • the information processing device 2 includes an element classification unit 207 that classifies each element constituting a sentence into a plurality of categories including at least a person category, and an element extraction unit 208 that extracts elements associated with the plurality of elements classified into the person category from among the elements constituting the plurality of sentences that have been structured.
  • the above configuration makes it possible to extract elements mentioned by multiple people.
  • Elements mentioned by multiple people are elements that should be focused on when investigating those people.
  • the information processing device 2 has the effect of facilitating an investigation targeting multiple people.
  • the classification of elements can be performed at any time after the sentences have been structured and the elements contained in the sentences have been extracted, and may be performed after S25 in FIG. 5, for example. If an input operation is performed to instruct the execution of element extraction when the elements have already been classified, S31 is omitted and the processes of S32 to S34 are performed.
  • the association between vehicles and people can be performed in advance by a user of the information processing device 2, and the results of the association can be input to the information processing device 2 via the communication unit 22 or the input unit 23.
  • the user can input the person to be investigated in advance to the information processing device 2, and can also input in advance to the information processing device 2 vehicles owned or used by that person, or vehicles suspected of being associated with that person.
  • the relevance identification unit 209 which has identified the time period during which the person traveled along the travel route, extracts elements of sentences structured by the structuring unit 203 that were sent or received during the identified time period by the person being investigated or by people who were accompanying the person. The relevance identification unit 209 then identifies, from among the extracted elements, elements that indicate a location as elements related to the research material.
  • the research materials may be, for example, a map including various information about locations such as place names, or materials showing various information about locations such as place names, store names, and addresses. Data detected using various search engines, etc. may also be used as the research materials.
  • the relevance identification unit 209 may associate search results related to locations obtained by searching using the identified element as a keyword with the element.
  • Figure 10 shows an example of the display of the results of identifying the correspondence between the elements of a structured sentence and the places shown in the research materials.
  • FIG. 10 shows a graph G4 indicating the results of structuring by the structuring unit 203, as well as an image 213b indicating the vehicle's movement route and each location on that movement route.
  • Image 213b is research material indicating each location on a specific vehicle's movement route detected by the vehicle detection system.
  • Image 213b indicates that the vehicle was detected at point p1 at 23:00 and at point p2 at 24:00, and also indicates that a parking area p3 and an interchange p4 are located on the route connecting points p1 and p2.
  • graph G4 the portion where the node for "Person A” and the node for "Place where you can park your car” are connected by an edge to "arrive” is generated from a sentence indicating that Person A will soon arrive at a place where he can park his car.
  • the character string "Place where you can park your car” is highlighted by being enclosed in a dashed line, and is displayed in association with "p3: Parking area” in image 213b by being linked with a dashed line.
  • the above portion of graph G4 also displays that the time when the above sentence was sent was 21:10. This allows the user to easily deduce that "Place where you can park your car” in the above sentence is parking area p3.
  • the materials 213 may include research materials showing a call history of a specific person.
  • the relation identification unit 209 may identify, as a sentence related to the call, a sentence sent or received by the person around the time when the research materials show that the person made a call.
  • text messages related to the phone call may be sent and received.
  • Such text messages may indicate the contents of the phone call either explicitly or implicitly. Therefore, according to the above configuration, in addition to the effects of the information processing device 1 according to the exemplary embodiment 1, the effect of facilitating the task of inferring the contents of the phone call from text messages in the research materials can be obtained.
  • the relation identification unit 209 first acquires research materials showing the call history of a specific person, and identifies the time period during which the call was made by that person. Next, the relation identification unit 209 identifies sentences sent or received by the person before or after the identified time period from among the sentences included in the target data 212. This associates the call history of the specific person with the sentences sent or received by that person before or after the call time period.
  • Figure 11 shows an example of the display of the results of identifying the correspondence between the structured sentences and the call history shown in the research materials.
  • graph G5 the part where the node with "Person A” and the node with “Person F” are connected by an edge with “Deposit” is generated from the message "Deposit within three days" sent by Person A to Person F.
  • This part is highlighted by being enclosed in a dashed line, and is displayed in association with the call history for April 30, 2023 in image 213c by being linked with a dashed line.
  • the above part of graph G5 also shows that the date and time when the above message was sent was 21:40 on April 30, 2023. This makes it easy to deduce that the content of Person A's call on April 30, 2023 was about the transfer of money.
  • the materials 213 may include investigation materials showing the deposit and withdrawal history of a specified account.
  • the association identifying unit 209 may identify the correspondence between the sentences and the investigation materials by comparing the times of deposits and withdrawals shown in the investigation materials with the times when sentences related to monetary transactions among the multiple sentences included in the target data 212 were sent or received.
  • Figure 12 shows an example of the display of the results of identifying the correspondence between the structured sentences and the deposit and withdrawal history shown in the research materials.
  • graph G6 the portion where the node for "Person F" and the node for "Payment of Compensation” are connected by an edge with “Necessary” was generated from a statement to the effect that Person F needs to pay compensation. This portion is highlighted by being enclosed in a dashed line, and is displayed in association with the payment history for May 1, 2023 in image 213d by linking it with the said payment history by a dashed line.
  • the above portion of graph G6 also shows that the date and time when the above statement was sent is 14:20 on May 1, 2023. This provides factual support that the payment on May 1, 2023 was prompted by the above statement sent at 14:20 on the same day.
  • the information processing device 2 can operate in cooperation with various external systems, not limited to a vehicle detection system. By cooperating with an external system, the information processing device 2 can acquire materials 213 collected by the system to be coordinated, and associate the acquired materials 213 with sentences or elements thereof included in the target data 212. The information processing device 2 may also acquire all or a part of the target data 212 from the external system.
  • the information processing device 2 that links with an external system can also be considered as part of that system, and various systems that include the information processing device 2 are included in the scope of the present invention.
  • various systems that include the information processing device 2 are included in the scope of the present invention.
  • a traffic control system that includes the information processing device 2 a monitoring system that includes the information processing device 2
  • a security system that includes the information processing device 2 are also included in the scope of the present invention.
  • the execution entity of each process described in the above embodiment is arbitrary and is not limited to the above example.
  • the functions of information processing devices 1 and 2 can be realized by multiple devices (which can also be called processors) that can communicate with each other.
  • processors which can also be called processors
  • each process described in the flow charts of Figures 2, 5, 7, and 8 can be shared and executed by multiple processors.
  • the execution entity of the structuring method in the above embodiment may be one processor or multiple processors.
  • An information processing device includes an element classification means for classifying each element in a plurality of structured sentences into a plurality of categories including at least a person category, and an element extraction means for extracting an element associated with the plurality of elements classified into the person category from among the elements.
  • This configuration has the effect of facilitating a survey targeting a plurality of people.
  • the above structuring may be performed by any method. In addition, it is not essential to infer the subject or object of a sentence when performing the structuring.
  • An information processing device includes a relationship identification means for identifying a correspondence between a plurality of sentences or elements thereof and a specified document based on the result of structuring the plurality of sentences, and an information presentation means for presenting information indicating the identification result of the relationship identification means.
  • This configuration has the effect of making it possible to easily recognize the correspondence between the sentences and the document.
  • the above structuring may be performed by any method. Furthermore, it is not essential to infer the subject or object of the sentence when performing the structuring.
  • Some or all of the functions of the information processing devices 1 and 2 may be realized by hardware such as an integrated circuit (IC chip), or may be realized by software.
  • the information processing devices 1 and 2 are realized, for example, by a computer that executes instructions of a program, which is software that realizes each function.
  • a computer that executes instructions of a program, which is software that realizes each function.
  • An example of such a computer (hereinafter referred to as computer C) is shown in Figure 13.
  • Computer C has at least one processor C1 and at least one memory C2.
  • Memory C2 stores a program (structured program) P for operating computer C as information processing device 1 or 2.
  • processor C1 reads and executes program P from memory C2, thereby realizing each function of information processing device 1 or 2.
  • the processor C1 may be, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating point number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination of these.
  • the memory C2 may be, for example, a flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination of these.
  • Computer C may further include a RAM (Random Access Memory) for expanding program P during execution and for temporarily storing various data.
  • Computer C may further include a communications interface for sending and receiving data to and from other devices.
  • Computer C may further include an input/output interface for connecting input/output devices such as a keyboard, mouse, display, and printer.
  • the program P can also be recorded on a non-transitory, tangible recording medium M that can be read by the computer C.
  • a recording medium M can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit.
  • the computer C can obtain the program P via such a recording medium M.
  • the program P can also be transmitted via a transmission medium.
  • a transmission medium can be, for example, a communications network or broadcast waves.
  • the computer C can also obtain the program P via such a transmission medium.
  • An information processing device comprising: a classification means for classifying sentences according to their contents; an estimation means for estimating at least one of a subject and an object of the sentence by applying an estimation method according to a result of the classification; and a structuring means for structuring the sentence based on a result of the estimation.
  • Appendix 2 The information processing device described in Appendix 1, wherein, for a sentence classified as an imperative sentence by the classification means, the estimation means performs at least one of a process of estimating a sender of the sentence to be a subject of the sentence and a process of estimating a recipient of the sentence to be an object of the sentence.
  • (Appendix 3) The information processing device according to claim 1, further comprising an element sequence extraction means for extracting a series of elements that are associated based on the result of the structuring and the result of classification by the classification means, from among the elements that constitute the plurality of sentences that have been structured.
  • Appendix 4 An information processing device as described in any one of appendices 1 to 3, comprising an element classification means for classifying each element constituting a sentence into a plurality of categories including at least a person category, and an element extraction means for extracting elements associated with the plurality of elements classified into the person category from among the elements constituting the plurality of sentences each of which has been structured.
  • Appendix 5 An information processing device as described in any of appendices 1 to 4, comprising: a relationship identification means for identifying a correspondence between the plurality of sentences or their elements and a specified material based on the result of structuring the plurality of sentences by the structuring means; and an information presentation means for presenting information indicating the result of the identification by the relationship identification means.
  • Appendix 6 The information processing device described in Appendix 5, wherein the specified materials include research materials showing the detection results of a specified vehicle by a vehicle detection system, and the relevance identification means identifies the correspondence between the sentences and the research materials by comparing the detection time of the vehicle shown in the research materials with the time when a person associated with the vehicle sent or received any of the sentences.
  • Appendix 7 The information processing device described in Appendix 5 or 6, wherein the specified materials include research materials indicating locations around a specified vehicle detected by a vehicle detection system or locations on the vehicle's travel route, and the relevance identification means identifies a correspondence between an element indicating a location included in a sentence sent or received by a person associated with the vehicle during the time period when the vehicle moved on the travel route or the time period when the vehicle was detected by the vehicle detection system, and the locations indicated in the research materials.
  • Appendix 8 An information processing device as described in any of Appendices 5 to 7, wherein the specified materials include investigation materials showing the call history of a specified person, and the relevance identification means identifies, as a sentence related to the call, a sentence sent or received by the person before or after a time period indicated in the investigation materials that the person made a call.
  • Appendix 9 An information processing device as described in any of Appendices 5 to 8, wherein the specified materials include investigation materials showing the deposit and withdrawal history of a specified account, and the correlation identification means identifies the correspondence between the sentences and the investigation materials by comparing the times of deposits and withdrawals shown in the investigation materials with the times when sentences related to the transfer of money among the multiple sentences were sent or received.
  • a structuring method comprising: at least one processor classifying sentences according to their contents; inferring at least one of a subject and an object of the sentences by applying a predetermined inference method according to a result of the classification; and structuring the sentences based on a result of the inference.
  • a computer-readable recording medium having recorded thereon a structuring program that causes a computer to function as a classification means for classifying sentences according to their contents, an estimation means for estimating at least one of the subject and object of the sentences by applying a predetermined estimation method according to the results of the classification, and a structuring means for structuring the sentences based on the results of the estimation.
  • the information processing device includes a relationship identification means for identifying a correspondence between a plurality of sentences or their elements and a specified material based on the result of structuring the plurality of sentences, and an information presentation means for presenting information indicating the identification result of the relationship identification means.
  • An information processing device comprising at least one processor that executes a process of classifying a sentence according to its content, a process of inferring at least one of a subject and an object of the sentence by applying an inference method according to a result of the classification, and a process of structuring the sentence based on a result of the inference.
  • the information processing device may further include a memory, and the memory may store a structuring program for causing the processor to execute the classification process, the estimation process, and the structuring process.
  • the structuring program may also be recorded on a computer-readable, non-transitory, tangible recording medium.
  • Appendix 2 The information processing device described in Appendix 1, wherein the at least one processor performs at least one of a process of inferring the sender of a sentence classified as an imperative sentence in the classification process as the subject of the sentence and a process of inferring the recipient of the sentence as the object of the sentence.
  • Appendix 3 The information processing device described in Appendix 1 or 2, wherein the at least one processor further performs a process of extracting a series of elements that are associated based on the result of the structuring and the result of classification by the classification means, from among the elements that constitute the multiple sentences that have been structured.
  • Appendix 4 An information processing device as described in any of appendices 1 to 3, wherein the at least one processor further executes an element classification means for classifying each element constituting a sentence into a plurality of categories including at least a person category, and a process for extracting elements associated with the plurality of elements classified into the person category from among the elements constituting the plurality of sentences each of which has been structured.
  • Appendix 5 An information processing device described in any of Appendices 1 to 4, wherein the at least one processor further executes a process of identifying a correspondence between the multiple sentences or their elements and a specified material based on the result of structuring the multiple sentences by the structuring means, and a process of presenting information indicating the identification result of the process.
  • Appendix 7 The information processing device described in Appendix 5 or 6, wherein the specified materials include research materials indicating locations around a specified vehicle detected by a vehicle detection system or locations on the vehicle's travel path, and the at least one processor identifies a correspondence between locations indicated in the research materials and elements included in a sentence sent or received by a person associated with the vehicle during the time period when the vehicle moved on the travel path or the time period when the vehicle was detected by the vehicle detection system.
  • Appendix 8 An information processing device as described in any of Appendix 5 to 7, wherein the specified materials include research materials showing the call history of a specified person, and the at least one processor identifies sentences sent or received by the person before or after a time period during which the research materials show that the person made a call as sentences related to the call.
  • Appendix 9 An information processing device as described in any of Appendices 5 to 8, wherein the specified materials include investigation materials showing the deposit and withdrawal history of a specified account, and the at least one processor identifies a correspondence between the sentences and the investigation materials by comparing the times of deposits and withdrawals shown in the investigation materials with the times when sentences related to the transfer of money among the multiple sentences were sent or received.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

Afin d'obtenir un résultat de structuration valide d'une phrase à qui manque au moins un élément parmi un sujet et/ou un objet, ce dispositif de traitement d'informations (1) comprend : une unité de classification (11) pour classifier une phrase en fonction de son contenu ; une unité d'estimation (12) pour appliquer un procédé d'estimation correspondant au résultat de la classification et estimer au moins un élément parmi le sujet et l'objet de la phrase ; et une unité de structuration (13) pour structurer la phrase sur la base du résultat de l'estimation.
PCT/JP2023/020951 2023-06-06 2023-06-06 Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement Pending WO2024252512A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/JP2023/020951 WO2024252512A1 (fr) 2023-06-06 2023-06-06 Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2023/020951 WO2024252512A1 (fr) 2023-06-06 2023-06-06 Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement

Publications (1)

Publication Number Publication Date
WO2024252512A1 true WO2024252512A1 (fr) 2024-12-12

Family

ID=93795353

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2023/020951 Pending WO2024252512A1 (fr) 2023-06-06 2023-06-06 Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement

Country Status (1)

Country Link
WO (1) WO2024252512A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007287134A (ja) * 2006-03-20 2007-11-01 Ricoh Co Ltd 情報抽出装置、及び情報抽出方法
JP2008077250A (ja) * 2006-09-19 2008-04-03 Ricoh Co Ltd 情報処理方法、情報処理装置、プログラム及びこれを記録した記録媒体
JP2012181685A (ja) * 2011-03-01 2012-09-20 Toshiba Corp 代表文抽出装置およびプログラム
WO2012127968A1 (fr) * 2011-03-23 2012-09-27 日本電気株式会社 Dispositif d'analyse d'événement, procédé d'analyse d'événement et support d'enregistrement lisible par ordinateur

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007287134A (ja) * 2006-03-20 2007-11-01 Ricoh Co Ltd 情報抽出装置、及び情報抽出方法
JP2008077250A (ja) * 2006-09-19 2008-04-03 Ricoh Co Ltd 情報処理方法、情報処理装置、プログラム及びこれを記録した記録媒体
JP2012181685A (ja) * 2011-03-01 2012-09-20 Toshiba Corp 代表文抽出装置およびプログラム
WO2012127968A1 (fr) * 2011-03-23 2012-09-27 日本電気株式会社 Dispositif d'analyse d'événement, procédé d'analyse d'événement et support d'enregistrement lisible par ordinateur

Similar Documents

Publication Publication Date Title
US11977995B2 (en) Machine learning artificial intelligence system for predicting hours of operation
US20220114676A1 (en) Methods and systems for automatically detecting fraud and compliance issues in expense reports and invoices
CN108133013B (zh) 信息处理方法、装置、计算机设备和存储介质
CN106407999A (zh) 结合规则来进行机器学习的方法及系统
Hajipour et al. A value-oriented Artificial Intelligence-as-a-Service business plan using integrated tools and services
WO2019196552A1 (fr) Procédé, appareil et dispositif de traitement de données permettant l'identification d'une fraude à l'assurance et serveur
TW202001736A (zh) 分類模型的訓練方法、店鋪分類的方法及裝置
US20230134796A1 (en) Named entity recognition system for sentiment labeling
CN109213859A (zh) 一种文本检测方法、装置及系统
US12333838B2 (en) Machine learning based information extraction
JP2015225362A (ja) 分析装置および分析方法
CN109582788A (zh) 垃圾评论训练、识别方法、装置、设备及可读存储介质
Averina et al. Impact of digital technologies on the company’s business model
US20190171977A1 (en) Using Machine Learning System to Dynamically Process Events
Johansson et al. Customer segmentation using machine learning
KR20220103015A (ko) 창업을 위한 정보를 제공하기 위한 전자 장치 및 그 동작 방법
US10762089B2 (en) Open ended question identification for investigations
US20220083570A1 (en) Enhanced data driven intelligent cloud advisor system
CN112200402B (zh) 一种基于风险画像的风险量化方法、装置及设备
CN117435471A (zh) 测试案例的推荐方法、装置、设备、存储介质和程序产品
CN113837772B (zh) 一种对营销信息进行审核的方法、装置及设备
CN115617998A (zh) 一种基于智能营销场景的文本分类方法及装置
CN114861680A (zh) 对话处理方法及装置
US20240403604A1 (en) Automatic Tokenization of Features Using Machine Learning
WO2024252512A1 (fr) Dispositif de traitement d'informations, procédé de structuration, et support d'enregistrement

Legal Events

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

Ref document number: 23940618

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2025525486

Country of ref document: JP

Kind code of ref document: A