Information processing device, information processing method and information processing program

The information processing device uses a binary classifier model to automatically differentiate between ambiguous and unambiguous domain-specific words, enhancing the accuracy and speed of natural language processing systems by constructing a dictionary through machine learning.

JP2024165576AActive Publication Date: 2024-11-28LY CORP
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2023081869
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-17
Publication Date
2024-11-28
Estimated Expiration
2043-05-17

AI Technical Summary

Technical Problem

Existing technologies fail to automatically distinguish between ambiguous and unambiguous domain-specific words, which are crucial for precise natural language processing, particularly in systems requiring high accuracy and speed.

Method used

An information processing device and method that utilizes a binary classifier model trained on co-occurrence data to identify unambiguous domain-specific words, constructing a dictionary that distinguishes between ambiguous and unambiguous proper nouns through machine learning techniques, including unsupervised and supervised learning phases.

Benefits of technology

Enables the automatic construction of a dictionary for unambiguous domain-specific words, improving the precision and speed of natural language processing systems, especially in named entity extraction and document classification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024165576000001_ABST
    Figure 2024165576000001_ABST
Patent Text Reader

Abstract

To automatically build a dictionary for unambiguous domain-specific words.SOLUTION: An information processing device according to the present application comprises: a learning unit that uses unambiguous proper nouns that are uniquely used with an intention of indicating a specific object and character strings that are related to the unambiguous proper nouns to build a model that estimates whether an input proper noun is an unambiguous proper noun through learning; a detection unit that inputs a proper noun to the learnt model, and detects the unambiguous proper noun; and a dictionary generation unit that automatically generates an unambiguous proper noun dictionary on the basis of the detected unambiguous proper nouns.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program. [Background technology]

[0002] A technique for automatically generating a dictionary has been disclosed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7261333 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the above conventional technology merely automatically generates a nursing dictionary used to support input into nursing records. Domain-specific words that indicate proper nouns in a specific domain (specific field or area) include ambiguous domain-specific words that are used not only as a base name that indicates a point on a map but also in other meanings, and non-ambiguous domain-specific words that are used only as a base name that indicates a point on a map. Conventional technologies could not automatically build a dictionary that distinguishes between ambiguous domain-specific words and non-ambiguous domain-specific words.

[0005] The present application has been made in view of the above, and aims to automatically build a dictionary of non-ambiguous domain-specific terms. [Means for solving the problem]

[0006] The information processing device of the present application is characterized by comprising a learning unit that constructs a model that estimates whether an inputted proper noun is an unambiguous proper noun through learning using unambiguous proper nouns that are uniquely used with the intention of indicating a specific object and character strings that are related to the unambiguous proper nouns, a detection unit that inputs proper nouns to the trained model and detects unambiguous proper nouns, and a dictionary generation unit that automatically generates an unambiguous proper noun dictionary based on the detected unambiguous proper nouns. Effect of the Invention

[0007] According to one aspect of the embodiment, a dictionary of unambiguous domain-specific terms is automatically constructed. It is possible. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is an explanatory diagram showing an overview of an information processing method according to an embodiment. [Diagram 2] FIG. 2 is an explanatory diagram showing an overview of application to a named entity extraction system. [Diagram 3] FIG. 3 is an explanatory diagram showing an outline of application to a document classification system. [Figure 4] FIG. 4 is a diagram illustrating an example of the configuration of an information processing system according to the embodiment. [Diagram 5] FIG. 5 is a diagram illustrating an example of the configuration of a terminal device according to the embodiment. [Figure 6] FIG. 6 is a diagram illustrating an example of the configuration of a server device according to the embodiment. [Figure 7] FIG. 7 is a diagram illustrating an example of the user information database. [Figure 8] FIG. 8 is a diagram illustrating an example of the history information database. [Figure 9] FIG. 9 is a diagram illustrating an example of the non-ambiguous information database. [Figure 10] FIG. 10 is a flowchart showing a processing procedure according to the embodiment. [Figure 11] FIG. 11 is a diagram illustrating an example of a hardware configuration. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, the information processing device, the information processing method, and the information processing program according to the present application will be described in detail with reference to the drawings. Note that the information processing device, the information processing method, and the information processing program according to the present application are not limited to the embodiments. In addition, the same components in the following embodiments are denoted by the same reference numerals, and duplicated descriptions are omitted.

[0010] [1. Overview of information processing method] First, an overview of an information processing method performed by an information processing device according to an embodiment will be described with reference to Fig. 1. Fig. 1 is an explanatory diagram showing an overview of an information processing method according to an embodiment. Note that Fig. 1 describes an example in which a dictionary of non-ambiguous domain-specific words is automatically constructed.

[0011] 1, the information processing system 1 includes a terminal device 10 and a server device 100. The terminal device 10 and the server device 100 are connected to each other via a network N (see FIG. 4) in a wired or wireless manner so as to be able to communicate with each other. In this embodiment, the terminal device 10 cooperates with the server device 100.

[0012] The terminal device 10 is a smart device such as a smartphone or tablet terminal used by a user U, and is a mobile terminal device capable of communicating with any server device via a wireless communication network such as 5G (Generation) or LTE (Long Term Evolution). The terminal device 10 has a screen such as a liquid crystal display having a touch panel function, and accepts various operations on display data such as content, such as tapping, sliding, scrolling, etc., performed by the user U with a finger or a stylus. An operation performed on an area of ​​the screen where the content is displayed may be regarded as an operation on the content. The terminal device 10 may be not only a smart device, but also an information processing device such as a desktop PC (Personal Computer) or a notebook PC.

[0013] The server device 100 is an information processing device that cooperates with the terminal device 10 of each user U and provides each user U's terminal device 10 with API (Application Programming Interface) services for various applications (hereinafter, apps), etc., as well as various data, and is realized by a computer, a cloud system, etc.

[0014] The server device 100 may also be an information processing device that provides some kind of online Web service to the terminal device 10 of each user U. For example, the server device 100 may provide services such as Internet connection, search service, SNS (Social Networking Service), electronic commerce (EC), electronic payment, online games, online banking, online trading, hotel and ticket reservations, video and music distribution, news, maps, route search, route guidance, line information, operation information, and weather forecasts as Web services. In practice, the server device 100 may cooperate with various servers that provide the above-mentioned Web services, and may mediate the Web services or be responsible for processing the Web services.

[0015] The server device 100 can acquire user information about the user U. For example, the server device 100 acquires information about the attributes of the user U, such as the gender, age, and residential area of ​​the user U. The server device 100 then stores and manages the information about the attributes of the user U together with identification information (such as a user ID) indicating the user U.

[0016] The server device 100 also acquires various types of history information (log data) indicating the actions of the user U from the terminal device 10 of the user U or from various servers based on the user ID or the like. For example, the server device 100 acquires a location history, which is a history of the location and date and time of the user U, from the terminal device 10. The server device 100 also acquires a search history, which is a history of search queries input by the user U, from a search server (search engine). The server device 100 also acquires a browsing history, which is a history of content browsed by the user U, from a content server. The server device 100 also acquires a purchase history (payment history), which is a history of product purchases and payment processing by the user U, from an electronic commerce server or a payment processing server. The server device 100 may also acquire a listing history, which is a history of listings on the marketplace by the user U, and a sales history, from the electronic commerce server or the payment processing server. The server device 100 also acquires a posting history, which is a history of posts by the user U, from a posting server that provides a word-of-mouth posting service or an SNS server. The above-mentioned various servers and the like may be the server device 100 itself. That is, the server device 100 may function as the above-mentioned various servers and the like.

[0017] [1-1. Automatic generation of non-ambiguous domain-specific term dictionaries] This paper describes a method for constructing a non-ambiguous domain-specific word dictionary using co-occurrence information of search queries. A non-ambiguous domain-specific word is a word (proper noun) that unambiguously indicates a single concept in a specific domain (specific field or area), and can be used in various natural language processing (NLP) systems.

[0018] A non-ambiguous domain-specific dictionary is useful when building a named entity extractor for a certain domain, for example, when you want to quickly build a high-speed extractor that prioritizes precision, or when the context of target words such as tweets and search queries is poor and disambiguation is difficult. However, there are countless non-ambiguous domain-specific words, and new words are constantly appearing, making it difficult to build them manually. Therefore, there is a demand for automatically building a non-ambiguous domain-specific dictionary.

[0019] In this embodiment, a dictionary of non-ambiguous domain specific words (non-ambiguous landmark words) targeting base names is automatically constructed. For example, words that are names of real facilities but are also used as general words (e.g., "Daruma" and "Tulip") and words that are base names and proper nouns but can be used for purposes other than landmark purposes (e.g., "Koshien") are ambiguous domain specific words (ambiguous landmark words). In contrast, words that rarely appear in any other sense than as base names that represent a point on a map (e.g., "Kioi Tower") are non-ambiguous domain specific words (non-ambiguous landmark words).

[0020] In addition, the tendency of co-occurrence word queries in search queries is different between ambiguous domain eigenwords and non-ambiguous domain eigenwords. For example, ambiguous domain eigenwords co-occur with various words other than landmark intent (e.g., “Koshien directions”, “Koshien breaking news”, “Koshien winning school”). In contrast, non-ambiguous domain eigenwords co-occur with specific words in landmark intent (e.g., “Kioi Tower directions”, “Kioi Tower restaurant”, “Kioi Tower access”). Therefore, a machine learning model using the co-occurrence word information of the search query is constructed. In this embodiment, a search query is introduced and a model (binary classifier) ​​that performs binary classification of whether it is a non-ambiguous domain eigenword or not is constructed. For example, when the word “Kioi Tower” is input, a model (binary classifier) ​​that classifies it as a non-ambiguous domain eigenword (non-ambiguous domain eigenword) is constructed. Alternatively, when the word "Kioi Tower" is input together with co-occurring words such as "directions," "access," or "restaurant," a model (binary classifier) ​​is constructed to classify the words into non-ambiguous domain-specific words (non-ambiguous domain-specific words).

[0021] For example, as shown in FIG. 1, the server device 100 acquires non-ambiguous proper nouns (non-ambiguous domain proper nouns), which are proper nouns that are used uniquely with the intention of indicating a specific object (such as a base name), and ambiguous proper nouns (ambiguous domain proper nouns), which are proper nouns that can be used not only with the intention of indicating a specific object but also with other intentions (step S1).

[0022] Next, the server device 100 collects a search query input together with the unambiguous proper noun (unambiguous co-occurrence word query) and a search query input together with the ambiguous proper noun (ambiguous co-occurrence word query) (step S2).

[0023] Next, the server device 100 constructs a learning dataset using each of the pairs of the unambiguous proper nouns and the unambiguous co-occurrence word queries and the pairs of the ambiguous proper nouns and the ambiguous co-occurrence word queries (step S3).

[0024] In practice, the server device 100 may collect data on pairs of unambiguous proper nouns and unambiguous co-occurrence word queries, and data on pairs of ambiguous proper nouns and ambiguous co-occurrence word queries, from logs of a search server or the like.

[0025] Next, the server device 100 constructs a model (binary classifier) ​​through learning using a data set of pairs of unambiguous proper nouns and unambiguous co-occurrence word queries and a data set of pairs of ambiguous proper nouns and ambiguous co-occurrence word queries (step S4).

[0026] The learning may be learning using a Transformer. For example, learning may be performed using Bidirectional Encoder Representations from Transformers (BERT) as follows.

[0027] (1) Preliminary learning During pre-learning, the server device 100 performs unsupervised learning that does not use labels. Note that the server device 100 does not necessarily need to perform pre-learning if a pre-trained large-scale model such as BERT is not used. In other words, pre-learning is essential for learning a large-scale model such as BERT, but is not essential for learning a model other than the large-scale model.

[0028] (2) Main study (re-study) During the main learning, the server device 100 performs supervised learning using labels. Examples of the main learning include transfer learning and fine-tuning. For example, the server device 100 learns a binary classification model using teacher data containing proper nouns and labels indicating whether they are unambiguous or ambiguous. The main learning is always essential.

[0029] (A) Proposed method 1 The server device 100 learns using the search query text during pre-learning (versus using general text in normal cases). For example, the server device 100 performs unsupervised learning using the search query text without using labels during pre-learning. As a method of this unsupervised learning, similar to the pre-learning method of BERT using a pre-trained model, learning is performed by randomly hiding (masking) tokens (≒ words) in the text and having the model infer the hidden parts (masked parts). This learning is unsupervised learning that does not require the use of correct answer labels constructed manually, and supervised learning such as classifying whether it is ambiguous or non-ambiguous is not performed. By performing this unsupervised learning using the search query text instead of general text, it becomes possible to implicitly embed information on how domain-specific words co-occur in the search query in advance in the machine learning model. In this way, the classification performance after this learning is improved. In addition, at the time of inference, by inputting one proper noun (a proper noun that is unclear whether it is an unambiguous proper noun or an ambiguous proper noun, such as a facility name candidate) into the model, it is possible to classify whether it is an unambiguous proper noun or not.

[0030] (B) Proposed method 2 During the main learning, the server device 100 learns a pair of a proper noun and a search query (co-occurrence word) as input data (as opposed to using a proper noun alone as input data in normal cases). For example, the server device 100 inputs a pair of an unambiguous proper noun and an unambiguous co-occurrence word query, or a pair of an ambiguous proper noun and an ambiguous co-occurrence word query, and constructs a model that estimates whether the proper noun is an unambiguous proper noun or an ambiguous proper noun. A model of any architecture can be used. In this case, during inference, the server device 100 inputs a proper noun into the model and classifies it into either an unambiguous proper noun or an ambiguous proper noun. This is because the model is trained to estimate a co-occurrence word query that is likely to be input together from the input proper noun, and estimate whether the proper noun is an unambiguous proper noun or an ambiguous proper noun in consideration of the estimated co-occurrence word query. The server device 100 may input a proper noun and a co-occurrence word query into the model and classify it into either an unambiguous proper noun or an ambiguous proper noun.

[0031] At this time, the server device 100 may assign a label of "unambiguous" to a dataset of a pair of an unambiguous proper noun and an unambiguous co-occurrence word query among the learning datasets, and may assign a label of "ambiguous" to a dataset of a pair of an ambiguous proper noun and an ambiguous co-occurrence word query. The server device 100 may perform labeling by crowdsourcing. The label may be "Yes" indicating an unambiguous proper noun and "No" indicating an ambiguous proper noun. In other words, it is sufficient if the label can be identified as "unambiguous" or "ambiguous".

[0032] Next, the server device 100 detects (automatically extracts) unambiguous proper nouns that can identify specific landmarks using the trained model (binary classifier) ​​(step S5).

[0033] For example, the server device 100 may input a proper noun into a model (binary classifier) ​​and perform binary classification as to whether it is an unambiguous proper noun or not. Unambiguous proper nouns have a different tendency in co-occurrence word queries than ambiguous proper nouns. In this embodiment, the server device 100 inputs a proper noun into a model (binary classifier) ​​and classifies it into either an unambiguous proper noun or an ambiguous proper noun. The server device 100 may input a pair of a proper noun and a co-occurrence word query into the model (binary classifier) ​​and perform binary classification as to whether it is an unambiguous proper noun or not.

[0034] Next, the server device 100 automatically constructs (automatically generates) an unambiguous proper noun dictionary (unambiguous domain specific word dictionary) based on the detected unambiguous proper nouns (step S6).

[0035] In this embodiment, the landmark (location information) domain is described as an example of the specific domain, but the same applies to other specific domains, such as shopping, medical, and recipes.

[0036] Thus, in this embodiment, the server device 100 learns a model (binary classifier) ​​that estimates whether an input proper noun (keyword, etc.) is an unambiguous proper noun or an ambiguous proper noun, using unambiguous proper nouns that are used uniquely with the intention of indicating a specific object (such as a base name), ambiguous proper nouns that can be used not only with the intention of indicating a specific object but also for other intentions, and character strings (co-occurring words, etc.) that are related to unambiguous proper nouns or ambiguous proper nouns.

[0037] The proper nouns may be extracted from an easily available list of entity names (such as a public web page or a database). The proper nouns and search queries may be character strings that appear in a text document. The proper nouns and search queries may be character strings extracted from image data by image recognition, or may be character strings extracted from voice data by voice recognition. The proper nouns are not limited to domain-specific words of the landmark domain, but may also be proper nouns in other fields, such as medicine (vague disease names, drug names, medical device names, ...) and recipes (vague dish names, food names, seasoning names, ...).

[0038] The relevant character string may be, for example, a search query that is entered together with a proper noun, or may be a word that frequently appears together with the proper noun in an article or content.

[0039] The server device 100 then uses a model (binary classifier) ​​to detect (automatically extract) unambiguous proper nouns and builds an unambiguous proper noun dictionary. The server device 100 may also input a search query to be judged to a model (binary classifier) ​​that estimates whether an input proper noun is an unambiguous proper noun or an ambiguous proper noun, estimate whether the proper noun is an ambiguous proper noun or an unambiguous proper noun, and vary the search results or suggest keywords to be input together depending on the estimation result.

[0040] As an application example, the server device 100 can apply the above model (binary classifier) ​​and the unambiguous proper noun dictionary to an existing system. Fig. 2 is an explanatory diagram showing an overview of application to a named entity extraction system. Fig. 3 is an explanatory diagram showing an overview of application to a document classification system.

[0041] For example, as shown in Fig. 2A, the server device 100 can quickly build a high-precision, high-speed named entity extraction system by incorporating an unambiguous proper noun dictionary into the named entity extraction system. This named entity extraction system can also handle cases where the context is scarce.

[0042] 2B, the server device 100 may add the above-mentioned model (binary classifier) ​​as a filter function to an existing named entity extraction system. It is not necessary to change the existing named entity extraction system itself, and performance can be improved just by adding the model (binary classifier).

[0043] Furthermore, as shown in FIG. 3, in the document classification system, the server device 100 may add the above-mentioned model (binary classifier) ​​as a filter function for deleting ambiguous features after extracting features, or may delete ambiguous features using an unambiguous proper noun dictionary.

[0044] Although not shown, the server device 100 may use a model (binary classifier) ​​or an unambiguous proper noun dictionary to determine the query intent of a shopping search. For example, the server device 100 may use a model (binary classifier) ​​or an unambiguous proper noun dictionary to detect unambiguous product names and use only unambiguous product names for the determination.

[0045] [2. Example of information processing system configuration] Next, a configuration of an information processing system 1 including a server device 100 according to an embodiment will be described with reference to Fig. 4. Fig. 4 is a diagram showing an example of a configuration of the information processing system 1 according to an embodiment. As shown in Fig. 4, the information processing system 1 according to an embodiment includes a terminal device 10 and a server device 100. These various devices are connected to each other via a network N so as to be able to communicate with each other by wire or wirelessly. The network N is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network) such as the Internet.

[0046] Furthermore, the number of devices included in the information processing system 1 shown in Fig. 4 is not limited to that shown in the figure. For example, in Fig. 4, for the sake of simplicity, only one terminal device 10 is shown, but this is merely an example and is not limiting, and two or more devices may be included.

[0047] The terminal device 10 is an information processing device used by a user U. For example, the terminal device 10 is a smart device such as a smartphone or a tablet terminal, a mobile phone such as a feature phone, a personal computer (PC), a personal digital assistant (PDA), a game machine or AV device equipped with a communication function, an information appliance or digital appliance, a car navigation system, a wearable device such as a smart watch or a head-mounted display, smart glasses, etc. The terminal device 10 may also be a house or building, a car, a home appliance, an electronic device, etc. that is compatible with the Internet of Things (IOT).

[0048] In addition, the terminal device 10 can connect to the network N via a wireless communication network such as LTE (Long Term Evolution), 4G (4th Generation), or 5G (5th Generation: 5th generation mobile communication system), or via short-range wireless communication such as Bluetooth (registered trademark) or wireless LAN (Local Area Network), and communicate with the server device 100.

[0049] The server device 100 is, for example, a computer such as a PC or a blade server, or a mainframe or a workstation, etc. The server device 100 may be realized by cloud computing.

[0050] [3. Example of terminal device configuration] Next, the configuration of the terminal device 10 will be described with reference to Fig. 5. Fig. 5 is a diagram showing an example of the configuration of the terminal device 10. As shown in Fig. 5, the terminal device 10 includes a communication unit 11, a display unit 12, an input unit 13, a positioning unit 14, a sensor unit 20, a control unit 30 (controller), and a storage unit 40.

[0051] (Communications Department 11) The communication unit 11 is connected to a network N (see FIG. 4) by wire or wirelessly, and transmits and receives information to and from the server device 100 via the network N. For example, the communication unit 11 is realized by a NIC (Network Interface Card), an antenna, or the like.

[0052] (Display section 12) The display unit 12 is a display device that displays various information such as position information. For example, the display unit 12 is a liquid crystal display (LCD) or an organic electro-luminescent display (OLED). The display unit 12 is a touch panel display, but is not limited to this.

[0053] (Input section 13) The input unit 13 is an input device that accepts various operations from the user U. For example, the input unit 13 has buttons for inputting characters, numbers, and the like. The input unit 13 may be an input / output port (I / O port), a USB (Universal Serial Bus) port, or the like. In addition, when the display unit 12 is a touch panel display, a part of the display unit 12 functions as the input unit 13. In addition, the input unit 13 may be a microphone that accepts voice input from the user U, or the like. The microphone may be wireless.

[0054] (Positioning unit 14) The positioning unit 14 receives a signal (radio wave) transmitted from a satellite of a GPS (Global Positioning System), and acquires position information (e.g., latitude and longitude) indicating the current position of the terminal device 10, which is the device itself, based on the received signal. That is, the positioning unit 14 measures the position of the terminal device 10. Note that the GPS is merely an example of a GNSS (Global Navigation Satellite System).

[0055] The positioning unit 14 can measure the position by various methods other than GPS. For example, the positioning unit 14 may measure the position by using various communication functions of the terminal device 10 as described below as an auxiliary positioning means for position correction or the like.

[0056] (Wi-Fi positioning) For example, the positioning unit 14 uses a Wi-Fi (registered trademark) communication function of the terminal device 10 or a communication network provided by each communication company to measure the position of the terminal device 10. Specifically, the positioning unit 14 performs Wi-Fi communication or the like and measures the distance to a nearby base station or access point to measure the position of the terminal device 10.

[0057] (Beacon positioning) The positioning unit 14 may measure the position by using a Bluetooth (registered trademark) function of the terminal device 10. For example, the positioning unit 14 measures the position of the terminal device 10 by connecting to a beacon transmitter connected by the Bluetooth (registered trademark) function.

[0058] (geomagnetic positioning) The positioning unit 14 also measures the position of the terminal device 10 based on a geomagnetic pattern of a structure measured in advance and a geomagnetic sensor included in the terminal device 10.

[0059] (RFID positioning) Furthermore, for example, if the terminal device 10 has a function of an RFID (Radio Frequency Identification) tag equivalent to a contactless IC card used at station ticket gates, stores, etc., or has a function of reading an RFID tag, the location of use is recorded together with information on a payment or the like made by the terminal device 10. The positioning unit 14 may obtain such information to measure the location of the terminal device 10. Furthermore, the location may be measured by an optical sensor, an infrared sensor, or the like provided in the terminal device 10.

[0060] The positioning unit 14 may, if necessary, measure the position of the terminal device 10 using one or a combination of the above-mentioned positioning means.

[0061] (Sensor unit 20) The sensor unit 20 includes various sensors mounted on or connected to the terminal device 10. The connection may be wired or wireless. For example, the sensors may be detection devices other than the terminal device 10, such as wearable devices and wireless devices. In the example shown in FIG. 5, the sensor unit 20 includes an acceleration sensor 21, a gyro sensor 22, an air pressure sensor 23, a temperature sensor 24, a sound sensor 25, a light sensor 26, a magnetic sensor 27, and an image sensor (camera) 28.

[0062] The above-mentioned sensors 21 to 28 are merely examples and are not intended to be limiting. That is, the sensor unit 20 may be configured to include some of the sensors 21 to 28, or may include other sensors such as a humidity sensor in addition to or instead of the sensors 21 to 28.

[0063] The acceleration sensor 21 is, for example, a three-axis acceleration sensor, and detects physical movements of the terminal device 10, such as the moving direction, speed, and acceleration of the terminal device 10. The gyro sensor 22 detects physical movements of the terminal device 10, such as tilts in three axial directions, based on the angular velocity, etc. of the terminal device 10. The air pressure sensor 23 detects, for example, the air pressure around the terminal device 10.

[0064] Since the terminal device 10 includes the above-mentioned acceleration sensor 21, gyro sensor 22, and air pressure sensor 23, it is possible to measure the position of the terminal device 10 using a technique such as Pedestrian Dead-Reckoning (PDR) that uses these sensors 21 to 23. This makes it possible to obtain indoor position information that is difficult to obtain using a positioning system such as GPS.

[0065] For example, the number of steps, walking speed, and distance walked can be calculated by a pedometer using the acceleration sensor 21. In addition, the moving direction, line of sight direction, and body inclination of the user U can be known by using the gyro sensor 22. In addition, the altitude and floor number on which the terminal device 10 of the user U is located can be known from the air pressure detected by the air pressure sensor 23.

[0066] The air temperature sensor 24 detects, for example, the air temperature around the terminal device 10. The sound sensor 25 detects, for example, the sound around the terminal device 10. The light sensor 26 detects the illuminance around the terminal device 10. The magnetic sensor 27 detects, for example, the geomagnetism around the terminal device 10. The image sensor 28 captures an image around the terminal device 10.

[0067] The above-mentioned air pressure sensor 23, temperature sensor 24, sound sensor 25, light sensor 26, and image sensor 28 can detect the air pressure, temperature, sound, and illuminance, and capture images of the surroundings, thereby detecting the environment and situation around the terminal device 10. In addition, the accuracy of the location information of the terminal device 10 can be improved based on the environment and situation around the terminal device 10.

[0068] (Control unit 30) The control unit 30 includes, for example, a microcomputer having a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM, an input / output port, and various other circuits. The control unit 30 may also be configured with hardware such as an integrated circuit, for example, an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The control unit 30 includes a transmitting unit 31, a receiving unit 32, and a processing unit 33.

[0069] (Transmitter 31) The transmission unit 31 can transmit, for example, various information input by the user U using the input unit 13, various information detected by each sensor 21-28 mounted on or connected to the terminal device 10, and location information of the terminal device 10 measured by the positioning unit 14 to the server device 100 via the communication unit 11.

[0070] (Receiving unit 32) The receiving unit 32 can receive various types of information provided by the server device 100 and requests for various types of information from the server device 100 via the communication unit 11.

[0071] (Processing section 33) The processing unit 33 controls the entire terminal device 10, including the display unit 12. For example, the processing unit 33 can output various information transmitted by the transmission unit 31 and various information received from the server device 100 by the reception unit 32 to the display unit 12 for display.

[0072] (Storage unit 40) The storage unit 40 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk drive (HDD), a solid state drive (SSD), an optical disk, etc. Various programs and various data are stored in the storage unit 40.

[0073] [4. Example of Server Device Configuration] Next, the configuration of the server device 100 according to the embodiment will be described with reference to Fig. 6. Fig. 6 is a diagram showing an example of the configuration of the server device 100 according to the embodiment. As shown in Fig. 6, the server device 100 includes a communication unit 110, a storage unit 120, and a control unit 130.

[0074] (Communication unit 110) The communication unit 110 is realized by, for example, a network interface card (NIC), etc. The communication unit 110 is also connected to a network N (see FIG. 4) in a wired or wireless manner.

[0075] (Storage unit 120) The storage unit 120 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a HDD, an SSD, an optical disk, etc. As shown in FIG. 6, the storage unit 120 has a user information database 121, a history information database 122, and a non-ambiguous information database 123.

[0076] (User Information Database 121) The user information database 121 stores user information about the user U. For example, the user information database 121 stores various information such as the attributes of the user U. Fig. 7 is a diagram showing an example of the user information database 121. In the example shown in Fig. 7, the user information database 121 has items such as "User ID (Identifier)", "Age", "Gender", "Home", "Workplace", and "Interests".

[0077] The "user ID" indicates identification information for identifying the user U. The "user ID" may be the contact information of the user U (such as a telephone number or an email address), or may be identification information for identifying the terminal device 10 of the user U.

[0078] Furthermore, "age" indicates the age of user U identified by the user ID. Note that "age" may be information indicating the specific age of user U (e.g., 35 years old) or information indicating the generation of user U (e.g., 30s). Alternatively, "age" may be information indicating user U's date of birth or information indicating user U's generation (e.g., born in the 1980s). Furthermore, "gender" indicates the gender of user U identified by the user ID.

[0079] Furthermore, "home" indicates the location information of the home of user U identified by the user ID. In the example shown in Fig. 7, "home" is illustrated as an abstract code such as "LC11", but it may also be latitude and longitude information, etc. Furthermore, for example, "home" may also be a region name or an address.

[0080] Furthermore, "workplace" indicates location information of the workplace (school in the case of a student) of user U identified by the user ID. Note that in the example shown in Fig. 7, "workplace" is illustrated as an abstract code such as "LC12", but it may also be latitude and longitude information, etc. Furthermore, for example, "workplace" may also be the name of a region or an address.

[0081] Furthermore, "interests" indicate the interests of a user U identified by a user ID. In other words, "interests" indicate subjects in which a user U identified by a user ID is highly interested. For example, "interests" may be search queries (keywords) that a user U inputs into a search engine. Note that, although one "interest" is illustrated for each user U in the example shown in FIG. 7, there may be multiple "interests."

[0082] For example, in the example shown in FIG. 7, the age of user U identified by user ID "U1" is "20s" and the gender is "male." Furthermore, for example, the user U identified by user ID "U1" indicates that his / her home address is "LC11." Furthermore, for example, the user U identified by user ID "U1" indicates that his / her workplace is "LC12." Furthermore, for example, the user U identified by user ID "U1" indicates that he / she is interested in "sports."

[0083] 7, abstract values ​​such as "U1", "LC11", and "LC12" are used for illustration, but "U1", "LC11", and "LC12" are assumed to store information such as specific character strings and numerical values. In the following figures relating to other information, abstract values ​​may also be illustrated.

[0084] The user information database 121 may store various information according to the purpose, not limited to the above. For example, the user information database 121 may store various information related to the terminal device 10 of the user U. The user information database 121 may also store information related to the attributes of the user U, such as demographic attributes, psychographic attributes, geographic attributes, and behavioral attributes. For example, the user information database 121 may store information such as name, family structure, place of origin (hometown), occupation, job title, income, qualifications, type of residence (detached house, apartment, etc.), whether or not the user has a car, commuting time, commuting route, commuter pass section (station, line, etc.), frequently used station (other than the nearest station to home or workplace), extracurricular activities (location, time zone, etc.), hobbies, interests, and lifestyle.

[0085] (Historical Information Database 122) The history information database 122 stores various information related to history information (log data) indicating the behavior of the user U. Fig. 8 is a diagram showing an example of the history information database 122. In the example shown in Fig. 8, the history information database 122 has items such as "user ID", "location history", "search history", "browsing history", "purchase history", and "posting history".

[0086] "User ID" indicates identification information for identifying user U. "Location history" indicates location history, which is a history of user U's location and movements. "Search history" indicates search history, which is a history of search queries entered by user U. "Browse history" indicates browse history, which is a history of content viewed by user U. "Purchase history" indicates purchase history, which is a history of purchases made by user U. "Post history" indicates posting history, which is a history of posts made by user U. "Post history" may include questions regarding user U's possessions.

[0087] For example, in the example shown in Figure 8, user U identified by user ID "U1" moved as shown in "Location History #1," searched as shown in "Search History #1," viewed content as shown in "Browse History #1," purchased specified products etc. at specified stores etc. as shown in "Purchase History #1," and posted as shown in "Post History #1."

[0088] Here, in the example shown in Figure 8, abstract values ​​such as "U1," "Location History #1," "Search History #1," "Browse History #1," "Purchase History #1," and "Post History #1" are used for the illustration, however, specific information such as character strings and numbers is stored in "U1," "Location History #1," "Search History #1," "Browse History #1," "Purchase History #1," and "Post History #1."

[0089] The history information database 122 may store various information according to the purpose, without being limited to the above. For example, the history information database 122 may store the usage history of a specific service by the user U. The history information database 122 may also store the history of the user U's visit to a physical store or the history of the user U's visit to a facility. The history information database 122 may also store the payment history of the user U's payment (electronic payment) using the terminal device 10.

[0090] (Unambiguous Information Database 123) The unambiguous information database 123 stores various information related to history information (log data) indicating the behavior of the user U. Fig. 9 is a diagram showing an example of the unambiguous information database 123. In the example shown in Fig. 9, the unambiguous information database 123 has items such as "proper noun", "unambiguous", and "co-occurring word".

[0091] A "proper noun" refers to a proper noun (keyword, etc.) that indicates a specific object (such as the name of a base). Note that proper nouns include non-ambiguous proper nouns (non-ambiguous domain proper nouns), which are proper nouns that are used uniquely with the intention of indicating a specific object, and ambiguous proper nouns (ambiguous domain proper nouns), which are proper nouns that can be used not only with the intention of indicating a specific object but also with other intentions.

[0092] Also, "unambiguous" indicates whether a proper noun is unambiguous or not. For example, it indicates whether a proper noun is an unambiguous proper noun or an ambiguous proper noun.

[0093] Furthermore, the term "co-occurrence word" refers to a character string (co-occurrence word, etc.) that is related to a proper noun. For example, the co-occurrence word includes a search query input together with an unambiguous proper noun (unambiguous co-occurrence word query) and a search query input together with an ambiguous proper noun (ambiguous co-occurrence word query).

[0094] For example, in the example shown in FIG. 9, the proper noun "Kioi Tower" is unambiguous "Yes", which means that it is an unambiguous proper noun, and also has a relationship (co-occurs) with co-occurring words such as "directions, access, restaurant".

[0095] The unambiguous information database 123 may store various information according to the purpose, not limited to the above. For example, the unambiguous information database 123 may store information on the attributes or segments of the user U who has searched for a proper noun together with co-occurring words. The unambiguous information database 123 may also store information on a model (binary classifier) ​​that performs binary classification as to whether or not a proper noun is unambiguous, or on an unambiguous proper noun dictionary.

[0096] (Control unit 130) Returning to Fig. 6, the description will be continued. The control unit 130 is a controller, and is realized by, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or the like, executing various programs (corresponding to an example of an information processing program) stored in a storage device inside the server device 100 using a storage area such as a RAM as a working area. In the example shown in Fig. 6, the control unit 130 has an acquisition unit 131, a learning unit 132, a detection unit 133, a dictionary generation unit 134, and an application unit 135.

[0097] (Acquisition part 131) The acquisition unit 131 acquires a search query input by a user U (user). For example, when the user U inputs a search query into a search engine or the like to perform a keyword search, the acquisition unit 131 acquires the search query via the communication unit 110. That is, the acquisition unit 131 acquires, via the communication unit 110, a keyword input by the user U into a search box of a search engine, a site, or an app.

[0098] Furthermore, the acquisition unit 131 acquires user information regarding the user U via the communication unit 110. For example, the acquisition unit 131 acquires identification information (such as a user ID) indicating the user U, location information of the user U, attribute information of the user U, and the like from the terminal device 10 of the user U. Furthermore, the acquisition unit 131 may acquire identification information indicating the user U, attribute information of the user U, and the like when registering the user U. Then, the acquisition unit 131 registers the user information in the user information database 121 of the storage unit 120.

[0099] Furthermore, the acquisition unit 131 acquires various types of history information (log data) indicating the behavior of the user U via the communication unit 110. For example, the acquisition unit 131 acquires various types of history information indicating the behavior of the user U from the terminal device 10 of the user U, or from various servers based on a user ID or the like. Then, the acquisition unit 131 registers the various types of history information in the history information database 122 of the storage unit 120.

[0100] In this embodiment, the acquiring unit 131 acquires a proper noun (keyword, etc.) and a character string related to the proper noun (co-occurring word, etc.) via the communication unit 110. The proper noun may be extracted from an easily available entity name list (publicly available web page, database, etc.). The related character string may be, for example, a search query input together with the proper noun, or a word that often appears together with the proper noun in a paper or content. The proper noun and the related character string may be a character string that appears in a text document. The proper noun and the character string may be a character string extracted from image data by image recognition, or a character string extracted from voice data by voice recognition.

[0101] For example, the acquisition unit 131 acquires a non-ambiguous proper noun (non-ambiguous domain proper word) that is uniquely used with the intention of indicating a specific object (such as a base name) and a character string related to the non-ambiguous proper noun. In this case, the acquisition unit 131 may acquire a non-ambiguous proper noun that is uniquely used with the intention of indicating a specific object, an ambiguous proper noun (ambiguous domain proper word) that can be used not only with the intention of indicating a specific object but also with other intentions, and a character string related to the non-ambiguous proper noun or the ambiguous proper noun.

[0102] (Learning Section 132) The learning unit 132 uses unambiguous proper nouns that are used uniquely with the intention of indicating a specific object and character strings that are related to unambiguous proper nouns to construct a model that, through learning, estimates whether an input proper noun is an unambiguous proper noun or not.

[0103] For example, the learning unit 132 uses unambiguous proper nouns that are used uniquely with the intention of indicating a specific object, ambiguous proper nouns that can be used not only with the intention of indicating a specific object but also for other intentions, and character strings that are related to unambiguous proper nouns or ambiguous proper nouns to construct a model that estimates whether an input proper noun is an unambiguous proper noun or an ambiguous proper noun through learning.

[0104] In addition, the learning unit 132 uses pairs of unambiguous proper nouns and character strings related to the unambiguous proper nouns, and pairs of ambiguous proper nouns and character strings related to the ambiguous proper nouns, as learning datasets, and constructs a model (binary classifier) ​​that classifies input proper nouns into unambiguous proper nouns or ambiguous proper nouns through learning.

[0105] In addition, the learning unit 132 constructs a classifier as the above model that classifies proper nouns related to inputted location names into unambiguous landmark words or ambiguous landmark words, using unambiguous landmark words that are used uniquely with the intention of indicating location names (landmarks, etc.) as unambiguous proper nouns, ambiguous landmark words that can be used not only with the intention of indicating location names as ambiguous proper nouns but also for other intentions, and co-occurring words of unambiguous landmark words or ambiguous landmark words.

[0106] In addition, the learning unit 132 uses unambiguous proper nouns, ambiguous proper nouns, and a search query input together with an unambiguous proper noun or an ambiguous proper noun to construct a classifier as the above model, which classifies an input proper noun into an unambiguous proper noun or an ambiguous proper noun.

[0107] Furthermore, the learning unit 132 learns using a search query text during pre-learning (as opposed to using general text as usual). For example, the learning unit 132 performs unsupervised learning using a search query text without using labels as pre-learning. For example, the learning unit 132 performs learning by randomly hiding (masking) tokens (≒ words) in the text and having the model infer the hidden parts (masked parts), similar to the pre-learning method of BERT that uses a pre-trained model.

[0108] Furthermore, the learning unit 132 learns pairs of proper nouns and search queries (co-occurring words) as input data during main learning (as opposed to normal use of a single proper noun as input data). For example, the learning unit 132 inputs pairs of proper nouns and co-occurring words (pairs of unambiguous proper nouns and co-occurring words, or pairs of ambiguous proper nouns and co-occurring words) to the model as main learning, and learns to classify the pairs into either unambiguous proper nouns or ambiguous proper nouns.

[0109] Furthermore, the learning unit 132 performs unsupervised learning in which no labels are used during pre-learning, and performs supervised learning in which labels indicating non-ambiguity or ambiguity are used during main learning.

[0110] (Detection unit 133) The detection unit 133 inputs a proper noun to the trained model and detects (automatically extracts) an unambiguous proper noun. The detection unit 133 may input a character string related to the proper noun together with the proper noun to the trained model to detect the unambiguous proper noun.

[0111] (Dictionary generation unit 134) The dictionary generation unit 134 automatically generates an unambiguous proper noun dictionary based on the detected unambiguous proper nouns.

[0112] (Applicable part 135) The application unit 135 incorporates or adds the above model or unambiguous proper noun dictionary to an existing system, thereby applying the above model or unambiguous proper noun dictionary to the existing system without modifying the existing system, thereby enabling detection of unambiguous proper nouns in the existing system. The application unit 135 may also be a providing unit that provides the above model or unambiguous proper noun dictionary to an external device via the communication unit 110.

[0113] 5. Processing Procedure Next, a processing procedure by the server device 100 according to the embodiment will be described with reference to Fig. 10. Fig. 10 is a flowchart showing the processing procedure according to the embodiment. Note that the processing procedure shown below is repeatedly executed by the control unit 130 of the server device 100.

[0114] 10, for example, the acquisition unit 131 of the server device 100 acquires proper nouns and co-occurring words that co-occur with the proper nouns via the communication unit 110 (step S101). In this embodiment, the acquisition unit 131 acquires non-ambiguous proper nouns that are used uniquely with the intention of indicating a specific object, ambiguous proper nouns that can be used not only with the intention of indicating a specific object but also with other intentions, and co-occurring words that co-occur with the non-ambiguous proper nouns or ambiguous proper nouns.

[0115] Next, the learning unit 132 of the server device 100 performs pre-learning, unsupervised learning without labels using the search query text, in which tokens (≒ words) in the text are randomly hidden (masked) and the model is made to infer the hidden parts (masked parts) (step S102).

[0116] Next, as the main learning, the learning unit 132 of the server device 100 inputs pairs of proper nouns and co-occurring words (pairs of unambiguous proper nouns and co-occurring words, or pairs of ambiguous proper nouns and co-occurring words) into a model (binary classifier) ​​and learns to classify them as either unambiguous proper nouns or ambiguous proper nouns (step S103).

[0117] Next, the detection unit 133 of the server device 100 inputs the proper noun to the trained model (binary classifier) ​​and detects (automatically extracts) the unambiguous proper noun (step S104). At this time, the detection unit 133 may input co-occurring words together with the proper noun to the trained model to detect the unambiguous proper noun.

[0118] Next, the dictionary generating unit 134 of the server device 100 automatically generates an unambiguous proper noun dictionary based on the detected unambiguous proper nouns (step S105).

[0119] Next, the application unit 135 of the server device 100 incorporates or adds the above model (binary classifier) ​​or unambiguous proper noun dictionary to the existing system, thereby applying the above model or unambiguous proper noun dictionary to the existing system without modifying the existing system, thereby enabling detection of unambiguous proper nouns in the existing system (step S106).

[0120] 6. Modifications The above-described terminal device 10 and server device 100 may be implemented in various different forms other than the above-described embodiment. Therefore, the following describes modifications of the embodiment.

[0121] In the above embodiment, some or all of the processing executed by the server device 100 may actually be executed by the terminal device 10. For example, the processing may be completed in a stand-alone manner (by the terminal device 10 alone). In this case, the terminal device 10 is assumed to have the functions of the server device 100 in the above embodiment. Also, in the above embodiment, since the terminal device 10 works in conjunction with the server device 100, from the perspective of the user U, it appears that the processing of the server device 100 is also executed by the terminal device 10. In other words, from another perspective, it can be said that the terminal device 10 is equipped with the server device 100.

[0122] Furthermore, in the above embodiment, the server device 100 may input a proper noun to a model (binary classifier) ​​and then output a score indicating the degree of certainty that the proper noun is an unambiguous proper noun as the output of the model (binary classifier), rather than whether the proper noun is an unambiguous proper noun or not. The server device 100 may then automatically construct (automatically generate) an unambiguous proper noun dictionary (unambiguous domain specific word dictionary) based on proper nouns (unambiguous proper nouns) whose scores are equal to or greater than a threshold. The score may be, for example, the output of a softmax function. The threshold may be a specified value, or may be arbitrarily settable.

[0123] In the above embodiment, the server device 100 may separately construct a model for classifying an input proper noun into whether it is a non-ambiguous proper noun and a model for classifying an input proper noun into whether it is an ambiguous proper noun. In other words, the model for detecting non-ambiguous proper nouns and the model for detecting ambiguous proper nouns may be independent of each other.

[0124] In the above embodiment, the server device 100 may sample 3 million web search queries as a method for explicitly adding co-occurrence words to proper nouns during main learning (transfer learning / fine tuning), and may construct a co-occurrence word dictionary for each word by dividing each web search query into space-delimited pieces. At this time, up to 10 different co-occurrence words are randomly collected for one target word. Next, the co-occurrence words obtained for the target noun, which is the input to the model, are linked with a [SEP] token to generate input data with co-occurrence words (e.g., “Kioi Tower [SEP] Access [SEP] Directions [SEP] Restaurant”). The [SEP] token is a token that represents the break in the sentence [SEP]. In this way, by linking with the [SEP] token, BERT can distinguish between the target noun to be actually classified and the co-occurrence words, which are additional information. Note that the target word and the co-occurrence words may be linked simply with a space delimited piece, not with the [SEP] token.

[0125] Moreover, it has been empirically found that non-ambiguous domain-specific words (non-ambiguous landmark words) appear less frequently in web search queries than general nouns or words with meanings other than landmarks. Therefore, in the above embodiment, a rule may be created that defines a noun whose frequency in web search queries is equal to or lower than a certain level (a threshold value or lower) as a non-ambiguous landmark word. Note that the threshold value may be a value that provides the highest F1 value in the learning data.

[0126] Furthermore, many of the words that can be non-ambiguous domain specific words (non-ambiguous landmark words) have specific suffixes, such as "XX Hospital" and "XX Post Office." Therefore, in the above embodiment, a suffix dictionary that can be non-ambiguous domain specific words (non-ambiguous landmark words) may be constructed in advance, and a rule may be created that determines a target word as a non-ambiguous landmark word when the suffix of the target word matches with the dictionary. The suffix dictionary may be a collection of frequently occurring suffixes from among the words included in a manually created non-ambiguous landmark word dictionary.

[0127] 7. Effects As described above, the information processing device (terminal device 10 and server device 100) of the present application includes a learning unit 132 that constructs a model for predicting whether an inputted proper noun is an unambiguous proper noun through learning using unambiguous proper nouns that are uniquely used with the intention of indicating a specific object and character strings that are related to the unambiguous proper nouns, a detection unit 133 that inputs proper nouns to the trained model and detects unambiguous proper nouns, and a dictionary generation unit 134 that automatically generates an unambiguous proper noun dictionary based on the detected unambiguous proper nouns.

[0128] The learning unit 132 uses unambiguous proper nouns that are used uniquely with the intention of indicating a specific object, ambiguous proper nouns that can be used not only with the intention of indicating a specific object but also for other intentions, and character strings that are related to unambiguous proper nouns or ambiguous proper nouns to construct a model that estimates whether an inputted proper noun is an unambiguous proper noun or an ambiguous proper noun through learning.

[0129] The learning unit 132 uses pairs of unambiguous proper nouns and character strings related to the unambiguous proper nouns, and pairs of ambiguous proper nouns and character strings related to the ambiguous proper nouns, as learning data sets, and constructs a model that classifies input proper nouns into unambiguous proper nouns or ambiguous proper nouns through learning.

[0130] The learning unit 132 constructs a classifier as the above model, which classifies proper nouns related to inputted location names into unambiguous landmark words or ambiguous landmark words, using unambiguous landmark words that are uniquely used with the intention of indicating the location name as an unambiguous proper noun, ambiguous landmark words that can be used not only with the intention of indicating the location name as an ambiguous proper noun but also for other intentions, and co-occurring words of unambiguous landmark words or ambiguous landmark words.

[0131] The learning unit 132 uses unambiguous proper nouns, ambiguous proper nouns, and a search query input together with an unambiguous proper noun or an ambiguous proper noun to construct a classifier as the above model that classifies an input proper noun into an unambiguous proper noun or an ambiguous proper noun.

[0132] As pre-learning, the learning unit 132 performs unsupervised learning using a search query text, in which tokens in the text are randomly masked and the masked parts are made to be inferred by a model.

[0133] In the main learning, the learning unit 132 inputs pairs of proper nouns and co-occurring words into the model, and learns to classify the pairs into either unambiguous proper nouns or ambiguous proper nouns.

[0134] The learning unit 132 performs unsupervised learning in which no labels are used during pre-learning, and performs supervised learning in which labels indicating non-ambiguity or ambiguity are used during main learning.

[0135] The detection unit 133 inputs a proper noun and a character string related to the proper noun to the trained model, and detects an unambiguous proper noun.

[0136] In addition, the information processing device of the present application further includes an application unit 135 that applies the above-mentioned model or unambiguous proper noun dictionary to an existing system by incorporating or adding the above-mentioned model or unambiguous proper noun dictionary to the existing system without modifying the existing system, thereby enabling detection of unambiguous proper nouns in the existing system.

[0137] By performing any one or a combination of the above-described processes, the information processing device according to the present application can automatically build a dictionary of non-ambiguous domain-specific words.

[0138] [8. Hardware Configuration] Moreover, the terminal device 10 and the server device 100 according to the above-described embodiment are realized by a computer 1000 having a configuration as shown in Fig. 11, for example. The server device 100 will be described below as an example. Fig. 11 is a diagram showing an example of a hardware configuration. The computer 1000 is connected to an output device 1010 and an input device 1020, and has a configuration in which a calculation device 1030, a primary storage device 1040, a secondary storage device 1050, an output I / F (Interface) 1060, an input I / F 1070, and a network I / F 1080 are connected by a bus 1090.

[0139] The arithmetic device 1030 operates based on programs stored in the primary storage device 1040 and the secondary storage device 1050, programs read from the input device 1020, and the like, and executes various processes. The arithmetic device 1030 is realized by, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or the like.

[0140] The primary storage device 1040 is a memory device such as a RAM (Random Access Memory) that primarily stores data used by the arithmetic device 1030 for various calculations. The secondary storage device 1050 is a storage device in which data used by the arithmetic device 1030 for various calculations and various databases are registered, and is realized by a ROM (Read Only Memory), an HDD (Hard Disk Drive), an SSD (Solid State Drive), a flash memory, or the like. The secondary storage device 1050 may be an internal storage device or an external storage device. The secondary storage device 1050 may be a removable storage medium such as a USB (Universal Serial Bus) memory or an SD (Secure Digital) memory card. The secondary storage device 1050 may be a cloud storage (online storage), a NAS (Network Attached Storage), a file server, or the like.

[0141] The output I / F 1060 is an interface for transmitting information to be output to an output device 1010 that outputs various types of information, such as a display, a projector, a printer, etc., and is realized by a connector conforming to a standard such as USB (Universal Serial Bus), DVI (Digital Visual Interface), or HDMI (registered trademark) (High Definition Multimedia Interface). The input I / F 1070 is an interface for receiving information from various input devices 1020, such as a mouse, a keyboard, a keypad, a button, a scanner, etc., and is realized by a USB, etc.

[0142] Furthermore, the output I / F 1060 and the input I / F 1070 may be wirelessly connected to the output device 1010 and the input device 1020, respectively. That is, the output device 1010 and the input device 1020 may be wireless devices.

[0143] The output device 1010 and the input device 1020 may be integrated into one device, such as a touch panel. In this case, the output I / F 1060 and the input I / F 1070 may also be integrated into one device, such as an input / output I / F.

[0144] The input device 1020 may be a device that reads information from, for example, an optical recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a PD (Phase change rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory.

[0145] The network I / F 1080 receives data from other devices via the network N and sends it to the arithmetic device 1030, and also transmits data generated by the arithmetic device 1030 to other devices via the network N.

[0146] The arithmetic unit 1030 controls the output device 1010 and the input device 1020 via the output I / F 1060 and the input I / F 1070. For example, the arithmetic unit 1030 loads a program from the input device 1020 or the secondary storage device 1050 onto the primary storage device 1040 and executes the loaded program.

[0147] For example, when the computer 1000 functions as the server device 100, the arithmetic unit 1030 of the computer 1000 executes a program loaded onto the primary storage device 1040 to realize the functions of the control unit 130. The arithmetic unit 1030 of the computer 1000 may also load a program acquired from another device via the network I / F 1080 onto the primary storage device 1040 and execute the loaded program. The arithmetic unit 1030 of the computer 1000 may also cooperate with the other device via the network I / F 1080 and use functions and data of a program by calling them from another program of the other device.

[0148] [9. Others] Although the embodiments of the present application have been described above, the present invention is not limited to the contents of these embodiments. The above-described components include those that can be easily imagined by a person skilled in the art, those that are substantially the same, and those that are within the so-called equivalent range. Furthermore, the above-described components can be appropriately combined. Furthermore, various omissions, substitutions, or modifications of the components can be made without departing from the spirit of the above-described embodiments.

[0149] Furthermore, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically by a known method. In addition, the information including the processing procedures, specific names, various data and parameters shown in the above documents and drawings can be changed arbitrarily unless otherwise specified. For example, the various information shown in each drawing is not limited to the illustrated information.

[0150] In addition, each component of each device shown in the figure is a functional concept, and does not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or part of them can be functionally or physically distributed and integrated in any unit according to various loads, usage conditions, etc.

[0151] For example, the above-mentioned server device 100 may be realized by multiple server computers, and depending on the functions, the configuration can be flexibly changed, such as by calling an external platform using an API (Application Programming Interface) or network computing.

[0152] Furthermore, the above-described embodiments and modifications can be appropriately combined as long as the processing contents are not contradictory.

[0153] Furthermore, the above-mentioned "section, module, unit" can be read as "means" or "circuit", etc. For example, an acquisition section can be read as an acquisition means or an acquisition circuit. [Explanation of symbols]

[0154] 1. Information Processing Systems 10 Terminal Equipment 100 Server device 110 Communications Department 120 Storage section 121 User Information Database 122 History Information Database 123 Unambiguous Information Database 130 Control section 131 Acquisition Department 132 Learning Department 133 Detection Unit 134 Dictionary Generation Unit 135 Application section

Claims

1. a learning unit that uses an unambiguous proper noun that is uniquely used with the intention of indicating a specific object and a character string that is related to the unambiguous proper noun to build a model that estimates whether an input proper noun is an unambiguous proper noun through learning; a detection unit that inputs proper nouns to the trained model and detects unambiguous proper nouns; a dictionary generation unit that automatically generates an unambiguous proper noun dictionary based on the detected unambiguous proper nouns; An information processing device comprising:

2. The learning unit uses the unambiguous proper noun that is used uniquely with the intention of indicating a specific object, an ambiguous proper noun that can be used not only with the intention of indicating a specific object but also with other intentions, and a character string related to the unambiguous proper noun or the ambiguous proper noun to build a model that estimates whether an input proper noun is an unambiguous proper noun or an ambiguous proper noun through learning.

2. The information processing apparatus according to claim 1,

3. The learning unit uses a pair of the unambiguous proper noun and a character string related to the unambiguous proper noun, and a pair of the ambiguous proper noun and a character string related to the ambiguous proper noun as a learning data set, and constructs a model for classifying an input proper noun into an unambiguous proper noun or an ambiguous proper noun through learning.

3. The information processing apparatus according to claim 2.

4. The learning unit uses an unambiguous landmark word that is uniquely used with the intention of indicating a place name as the unambiguous proper noun, an ambiguous landmark word that can be used not only with the intention of indicating a place name as the ambiguous proper noun but also with other intentions, and co-occurring words of the unambiguous landmark word or the ambiguous landmark word to construct a classifier that classifies a proper noun related to an input place name into an unambiguous landmark word or an ambiguous landmark word as the model.

3. The information processing apparatus according to claim 2.

5. The learning unit uses the unambiguous proper noun, the ambiguous proper noun, and a search query input together with the unambiguous proper noun or the ambiguous proper noun to construct a classifier as the model for classifying an input proper noun into an unambiguous proper noun or an ambiguous proper noun.

3. The information processing apparatus according to claim 2.

6. The learning unit performs unsupervised learning using a search query text as pre-learning, in which tokens in the text are randomly masked and the masked parts are made to be inferred by a model.

3. The information processing apparatus according to claim 2.

7. The learning unit inputs pairs of proper nouns and co-occurring words to the model as main learning, and learns to classify the pairs into either unambiguous proper nouns or ambiguous proper nouns.

3. The information processing apparatus according to claim 2.

8. The learning unit performs unsupervised learning without using labels in pre-learning, and performs supervised learning using labels indicating non-ambiguous or ambiguous in main learning.

3. The information processing apparatus according to claim 2.

9. The detection unit detects an unambiguous proper noun by inputting a proper noun and a character string related to the proper noun to the trained model.

3. The information processing apparatus according to claim 1, wherein the information processing apparatus is a computer.

10. an application unit that applies the model or the unambiguous proper noun dictionary to an existing system by incorporating or adding the model or the unambiguous proper noun dictionary to the existing system without modifying the existing system, thereby enabling detection of unambiguous proper nouns in the existing system; 3. The information processing apparatus according to claim 1, further comprising:

11. An information processing method executed by an information processing device, A learning process for constructing a model that estimates whether an input proper noun is an unambiguous proper noun through learning using an unambiguous proper noun that is uniquely used with the intention of indicating a specific object and a character string that is related to the unambiguous proper noun; a detection step of inputting proper nouns to the trained model to detect unambiguous proper nouns; a dictionary generation step of automatically generating an unambiguous proper noun dictionary based on the detected unambiguous proper nouns; 13. An information processing method comprising:

12. A learning procedure for constructing a model that estimates whether an input proper noun is an unambiguous proper noun through learning using an unambiguous proper noun that is uniquely used with the intention of indicating a specific object and a character string that is related to the unambiguous proper noun; A detection step of inputting proper nouns to the trained model to detect unambiguous proper nouns; A dictionary generation procedure for automatically generating an unambiguous proper noun dictionary based on the detected unambiguous proper nouns; An information processing program characterized by causing a computer to execute the above.

Citation Information

Patent Citations

  • Specific expression judging device and specific expression judging method, and recording medium with specific expression judging program recorded thereon

    JP2002082943A

  • Specific expression extraction device and method

    JP2007148785A

  • Disambiguation method and system

    JP2010225135A

  • Entity linking method, device, electronic device, storage medium, and computer program

    JP2021168124A

  • Dictionary generation device, dictionary generation method, and dictionary generation program

    JP7261333B1