JP7453199B2 - Information processing device, information processing method, and information processing program - Google Patents

Description

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

Conventionally, entity linking is used to return related entities in response to a query. Entity linking is a general term for methods that return related entities using input text. For example, the word "Softbank (registered trademark)" may be associated with a "company" or a "team." In this way, since there are different entities with the same (similar) notation, it is important to reflect the frequency information and context information of the click log in the determination result.

Non-Patent Document 1 discloses a technology that returns entities with high probability in response to a query by introducing a probability model based on click logs. Non-Patent Document 2 proposes a method of identifying entities as main words and peripheral words by employing words included in the first sentence of Wikipedia (registered trademark) as a white list and using them as peripheral words.

Fast and Space-Efficient Entity Linking in Queries <URL:http://groups.di.unipi.it/~ottavian/files/wsdm15_fel.pdf> Unsupervised entity linking considering peripheral words for web search queries <URL: https://www.anlp.jp/proceedings/annual_meeting/2019/pdf_dir/F1-1.pdf>

However, with the above-mentioned conventional technology, it is not possible to determine when a word outside the whitelist appears as a peripheral word. For example, if you search with the query "Softbank (registered trademark) game," it is expected that search results related to "Fukuoka SoftBank Hawks (registered trademark)" (team) will be returned, but if "game" is included in the whitelist, This kind of determination is difficult if no words are included. Also, when multiple entities are linked to a query, or when two or more words are linked to one entity (for example, when the query "new game latest edition" is linked to the entity "new game (book)"), As the number of words in a query increases, it becomes difficult to make robust judgments.

The present application was made in view of the above, and aims to realize robust entity linking using context consistency.

The information processing device according to the present application includes a reception unit that receives input of a query, a calculation unit that constructs a lattice structure using candidate entities of the word sequence of the query, and calculates a maximum likelihood entity sequence, and a calculation unit that calculates the maximum likelihood entity sequence. a providing unit that provides search results according to the word sequence, and the calculation unit lists candidate entities with probabilities for the word series, calculates the relationship between the candidate entities by an inner product of virtual documents, and calculates the relationship between the candidate entities and the probabilities. The method is characterized in that the logarithmic sum of the inner product is calculated, virtual nodes are placed before and after the lattice, and a path with the maximum score between the virtual nodes is determined to obtain a maximum likelihood entity sequence .

According to one aspect of an embodiment, robust entity linking using context consistency can be achieved.

FIG. 1 is an explanatory diagram showing an overview of an information processing method according to an embodiment. FIG. 2 is a diagram showing an example of evaluation data for an evaluation experiment. FIG. 3 is a diagram illustrating a configuration example of an information processing system according to an embodiment. FIG. 4 is a diagram illustrating a configuration example of a terminal device according to an embodiment. FIG. 5 is a diagram illustrating a configuration example of an information providing device according to an embodiment. FIG. 6 is a diagram showing an example of a user information database. FIG. 7 is a diagram showing an example of a history information database. FIG. 8 is a diagram illustrating an example of a relevance information database. FIG. 9 is a flowchart showing the processing procedure according to the embodiment. FIG. 10 is a diagram showing an example of a hardware configuration.

DESCRIPTION OF THE PREFERRED EMBODIMENTS An information processing apparatus, an information processing method, and an information processing program according to the present application (hereinafter referred to as "embodiments") will be described in detail below with reference to the drawings. Note that the information processing apparatus, information processing method, and information processing program according to the present application are not limited to this embodiment. In addition, in the following embodiments, the same parts are given the same reference numerals, and redundant explanations will be omitted.

[1. Overview of information processing method]
First, with reference to FIG. 1, an overview of an information processing method performed by an information processing apparatus according to an embodiment will be described. FIG. 1 is an explanatory diagram showing an overview of an information processing method according to an embodiment. Note that in FIG. 1, an example will be described in which robust entity linking using context consistency is implemented.

As shown in FIG. 1, the information processing system 1 includes a terminal device 10 and an information providing device 100. The terminal device 10 and the information providing device 100 are connected to be able to communicate with each other by wire or wirelessly via a network N (see FIG. 3). In this embodiment, the terminal device 10 cooperates with the information providing device 100.

The terminal device 10 is a smart device such as a smartphone or a tablet used by a user U (user), and can communicate with any server device via a wireless communication network such as 4G (Generation) or LTE (Long Term Evolution). This is a mobile terminal device that can perform Further, the terminal device 10 has a screen such as a liquid crystal display, which has a touch panel function, and displays data such as content through a tap operation, a slide operation, a scroll operation, etc. from the user U using a finger or a stylus. Accepts various operations on. Note that an operation performed on an area of the screen where content is displayed may be an operation on the content. Further, 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.

The information providing device 100 cooperates with the terminal device 10 of each user U, and provides API (Application Programming Interface) services for various applications (hereinafter referred to as applications), etc., to the terminal device 10 of each user U, and various It is an information processing device that provides data, and is realized by a server device, a cloud system, etc.

Further, the information providing device 100 may be an information processing device that provides some kind of web service online to the terminal device 10 of each user U. For example, the information providing device 100 provides web services such as Internet connection, search service, SNS (Social Networking Service), electronic commerce (EC), electronic payment, online games, online banking, online trading, accommodation and ticket reservations. , video/music distribution, news, maps, route searches, route guidance, route information, operation information, weather forecasts, and other services may be provided. In reality, the information providing apparatus 100 may cooperate with various servers that provide the above-mentioned Web services, mediate the Web services, or may be in charge of processing the Web services.

Note that the information providing device 100 can acquire user information regarding the user U. For example, the information providing device 100 acquires information regarding attributes of the user U, such as the user's U's gender, age, and area of residence. Then, the information providing device 100 stores and manages information regarding attributes of the user U along with identification information indicating the user U (user ID, etc.).

Further, the information providing device 100 acquires various kinds of history information (log data) indicating the behavior 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 information providing device 100 acquires a location history, which is a history of the user U's location, date and time, from the terminal device 10. The information providing 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 information providing device 100 also acquires a viewing history, which is a history of content viewed by the user U, from the content server. Further, the information providing device 100 acquires a purchase history (payment history) that is a history of product purchases and payment processing by the user U from the electronic commerce server and the payment processing server. Further, the information providing device 100 may acquire the listing history and the sales history, which are the listing history of the user U on the marketplace, from the electronic commerce server or the payment server. The information providing device 100 also acquires a posting history, which is a history of postings by the user U, from a posting server or SNS server that provides a word-of-mouth posting service.

[1-1. Objective function]
In this embodiment, the information providing apparatus 100 extends the framework of morphological analysis to entity linking. For example, the information providing device 100 obtains the entity sequence Y with the maximum score for the word sequence X. Equation (1) below is a formula for finding an entity sequence Y that has the maximum probability when a word sequence X is given. In Equation (1) below, (y _t , y _t-1 ) represents that the current entity is influenced only by the entity immediately before it. K represents the magnification of how strongly the transition feature is reflected relative to the observed feature.

(observation feature)
The information providing apparatus 100 uses, for example, a log probability sum maximization model of FEL (Fast Entity Linking), which is a method described in Non-Patent Document 1. For example, the information providing device 100 calculates the probability of occurrence of the entity e for the word sequence s using P(e|s) as shown in Equation (2) below.

(transition feature)
The information providing device 100 models an entity as an LDA (Latent Dirichlet Allocation) document generated from a clicked word. For example, as described in Non-Patent Document 2, the information providing apparatus 100 views a document as a virtual document composed of clicked words, and models the document using a vectorization method using LDA. As a result, each entity is assigned a vector having several hundred (eg, 500) dimensions. Note that LDA is an example of natural language processing. Actually, it may be word2vec or the like. Furthermore, the information providing apparatus 100 calculates the relationship between entities using the inner product of the entities. Then, the information providing device 100 constructs a lattice structure using the entity candidates of the word series, and obtains the most likely entity series. For example, the information providing device 100 estimates the route with the maximum score using the Viterbi algorithm as described above (see Equation 1).

[1-2. algorithm〕
As shown in FIG. 1, the information providing device 100 receives a query input from the terminal device 10 of each user U via the network N (see FIG. 3), and searches based on the query (step S1). .

Subsequently, the information providing apparatus 100 lists candidate entities along with their probabilities for the word sequence of the query (step S2).

Subsequently, the information providing apparatus 100 calculates the relationship between the candidate entities using the inner product of the virtual documents (step S3).

Subsequently, the information providing device 100 calculates the logarithmic sum of the probability and the inner product, since the score of the route is expressed by the accumulation of the probability and the inner product (step S4).

Subsequently, the information providing apparatus 100 installs virtual nodes before and after the lattice, finds a route with the maximum score between the virtual nodes, and obtains a maximum likelihood entity sequence (step S5).

Subsequently, the information providing device 100 returns a search result corresponding to the maximum likelihood entity sequence to the terminal device 10 of each user U via the network N (see FIG. 3) (step S6).

Specifically, the information providing apparatus 100 lists candidate entities along with probabilities for the word series (corresponding to nodes in FIG. 1). At this time, the information providing device 100 calculates the probability using P(e|s). Furthermore, the information providing apparatus 100 calculates the relationship between candidate entities using the inner product of the virtual document (corresponding to the edges in FIG. 1). Furthermore, the information providing device 100 calculates the logarithmic sum since the score of the route is expressed by the accumulation of probability (<=1) and inner product (<=1). The information providing device 100 also installs virtual nodes "start" and "end" before and after the lattice, and finds a route with the maximum score between "start" and "end."

At this time, the information providing apparatus 100 applies Japanese morphological analysis using a CRF (Conditional Random Field) to the word sequence of the query (obtains a maximum likelihood entity sequence). Furthermore, the information providing apparatus 100 estimates that a word that is not associated with an entity is a virtual document consisting of the word itself. Then, the information providing device 100 evaluates the overall consistency by treating entities and word sequences (non-entities) in a unified manner. Note that the information providing device 100 uses the hyperparameter β (ratio of probability and inner product) to calculate as “normalization term=number of nodes*1+number of edges*β”.

In the example in Figure 1, for the query "Kingdom Ken Suzuki", if the relationship between candidate entities is not taken into consideration, judgment will be made for each individual word, and "Kingdom" with the highest score for the word "Kingdom" will be judged. (Manga)'' is selected, the entity of "Ken Suzuki (Infielder)" with the highest score for the word "Ken Suzuki" is selected, and the entity of "Kingdom (Manga) Ken Suzuki (Infielder)" is selected, A route is returned. In this embodiment, by considering the relationship between candidate entities, a route for "Kingdom (pro-wrestling) Ken Suzuki (pro-wrestling reporter)" is returned in response to the query "Kingdom Ken Suzuki".

Furthermore, the information providing device 100 calculates a vector representation of the word sequence and evaluates the consistency. For example, when there is an entity associated with a word series in a KB (Knowledge Base), the information providing device 100 calculates a topic vector of a virtual document consisting of all click logs associated with the entity and checks the consistency. evaluate. Furthermore, when there is no entity associated with a word series in the KB, the information providing apparatus 100 calculates a topic vector of a virtual document consisting of a single word and evaluates consistency. As the KB becomes more and more filled with entities, the accuracy increases (even if they do not exist, calculations can be made to some extent). Furthermore, the information providing device 100 identifies entities from word sequences of arbitrary length. Further, the information providing device 100 can identify not only words separated by blank spaces but also word strings.

For example, the information providing device 100 calculates the inner product of the occurrence probability of the book "new game" and the word vector "latest publication" in response to the query "new game latest edition", and calculates the inner product of the occurrence probability and the inner product. If the total value of is the highest score, then "new game (book) latest publication (word)" will be the most likely entity sequence for the query "new game latest publication". At this time, internally, the information providing device 100 separates "new game latest issue" with a blank space, and divides it into three patterns (first pattern): "new", "game", and "latest issue". A pattern in which "new game" and "latest edition" are divided into two and one (second pattern), a pattern in which "new" and "game latest edition" are divided into one and two (third pattern), and a pattern in which "new" and "latest edition" are divided into two and one, The calculation is made for each of the four patterns (fourth pattern) that are fixed in one pattern (the fourth pattern) of "Latest issue of game", and the one with the highest score is returned.

Further, the information providing device 100 may model a query as a sentence. At this time, the information providing apparatus 100 considers the relationship between adjacent words. For example, in the information providing apparatus 100, the user does not input words at random, but inputs words that seem to be related in order. Furthermore, even if the number of words increases, the information providing device 100 can perform calculations in O(n) using the Viterbi algorithm. For example, when modeled as a document, the amount of calculation for all combinations is O(n^2).

[1-3. Evaluation experiment]
(evaluation data)
FIG. 2 is a diagram showing an example of evaluation data for an evaluation experiment. As shown in FIG. 2, the information providing device 100 blank-connects the team parent company (12 cases) and the baseball context word (20 cases) to define an evaluation query (240 cases in total). For example, the information providing device 100 blankly connects the team parent company "Orix" and the baseball context "game" to define the evaluation query "Orix game." If the context is not taken into consideration, it is possible that the decision will be made not to the team but to the company or region (prefecture). As a result of the evaluation experiment, an improvement of about 13.8% was achieved among these evaluation queries (240 in total).

Although there are methods for morphological analysis and named entity extraction by treating a query as a sentence, existing methods only obtain the part of speech and named entity of a word, and there is no method for entity linking. Further, although elemental technologies such as morphological analysis and FEL are known, the combination as in this embodiment does not exist in the past and is new.

(Feature design)
In this embodiment, the information providing apparatus 100 is designed to use an observation feature (eg, FEL) and a transition feature (eg, topic vector inner product) as feature quantities. The feature amount is an example of an effective value, and it is possible to add it. Further, in the present embodiment, the information providing apparatus 100 performs vector representation of word sequences, detection of entities linked to word sequences of arbitrary length, and the like.

[2. Configuration example of information processing system]
Next, the configuration of the information processing system 1 including the information providing apparatus 100 according to the embodiment will be described using FIG. 3. FIG. 3 is a diagram showing a configuration example of the information processing system 1 according to the embodiment. As shown in FIG. 3, the information processing system 1 according to the embodiment includes a terminal device 10 and an information providing device 100. These various devices are connected via a network N so that they can communicate 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.

Further, the number of each device included in the information processing system 1 shown in FIG. 3 is not limited to what is illustrated. For example, in FIG. 3, only one terminal device 10 is shown for simplification of illustration, but this is just an example and is not limited, and there may be two or more terminal devices.

The terminal device 10 is an information processing device used by the user U. For example, the terminal device 10 may be a smart device such as a smartphone or a tablet terminal, a feature phone, a PC (Personal Computer), a PDA (Personal Digital Assistant), a game console or AV device with a communication function, a car navigation system, a smart watch, or the like. These include wearable devices such as head-mounted displays, smart glasses, etc.

The terminal device 10 also supports wireless communication networks such as LTE (Long Term Evolution), 4G (4th Generation), and 5G (5th Generation), Bluetooth (registered trademark), and wireless LAN (Local It is possible to connect to a network N via short-distance wireless communication such as a local area network) and communicate with the information providing device 100.

The information providing device 100 is, for example, a PC, a server device, a mainframe, a workstation, or the like. Note that the information providing device 100 may be realized by cloud computing.

[3. Configuration example of terminal device]
Next, the configuration of the terminal device 10 will be explained using FIG. 4. FIG. 4 is a diagram showing a configuration example of the terminal device 10. As shown in FIG. As shown in FIG. 4, the terminal device 10 includes a communication section 11, a display section 12, an input section 13, a positioning section 14, a sensor section 20, a control section 30 (controller), and a storage section 40. Be prepared.

(Communication Department 11)
The communication unit 11 is connected to a network N (see FIG. 3) by wire or wirelessly, and transmits and receives information to and from the information providing 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.

(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 (EL display). Further, the display unit 12 is a touch panel type display, but is not limited to this.

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

(Positioning unit 14)
The positioning unit 14 receives a signal (radio wave) sent from a GPS (Global Positioning System) satellite, and based on the received signal, determines position information (for example, latitude and longitude). That is, the positioning unit 14 positions the terminal device 10 . Note that GPS is just one example of GNSS (Global Navigation Satellite System).

Further, the positioning unit 14 can measure the position using various methods other than GPS. For example, the positioning unit 14 may use various communication functions of the terminal device 10 to measure the position as an auxiliary positioning means for position correction and the like, as described below.

(Wi-Fi positioning)
For example, the positioning unit 14 positions the terminal device 10 using the Wi-Fi (registered trademark) communication function of the terminal device 10 or the communication network provided by each communication company. Specifically, the positioning unit 14 performs Wi-Fi communication, etc., and determines the position of the terminal device 10 by measuring the distance to nearby base stations and access points.

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

(geomagnetic positioning)
Furthermore, the positioning unit 14 positions the terminal device 10 based on the geomagnetic pattern of the structure measured in advance and the geomagnetic sensor included in the terminal device 10 .

(RFID positioning)
Further, for example, if the terminal device 10 has an RFID (Radio Frequency Identification) tag function equivalent to a contactless IC card used at station ticket gates, stores, etc., or has a function to read an RFID tag. In this case, the location where the terminal device 10 used the terminal device 10 is recorded together with the information that the payment was made. The positioning unit 14 may measure the position of the terminal device 10 by acquiring such information. Further, the position may be determined by an optical sensor, an infrared sensor, or the like provided in the terminal device 10.

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

(sensor section 20)
The sensor unit 20 includes various sensors mounted on or connected to the terminal device 10. Note that the connection may be a wired connection or a wireless connection. For example, the sensors may be a detection device other than the terminal device 10, such as a wearable device or a wireless device. In the example shown in FIG. 4, the sensor unit 20 includes an acceleration sensor 21, a gyro sensor 22, an atmospheric pressure sensor 23, an air temperature sensor 24, a sound sensor 25, an optical sensor 26, a magnetic sensor 27, and an image sensor ( camera) 28.

Note that each of the sensors 21 to 28 described above is merely an example and is not limited to the above. That is, the sensor section 20 may be configured to include a portion of each of the sensors 21 to 28, or may include other sensors such as a humidity sensor in addition to or instead of each of the sensors 21 to 28. .

The acceleration sensor 21 is, for example, a three-axis acceleration sensor, and detects the physical movement 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 tilt in three axes directions based on the angular velocity of the terminal device 10 and the like. The atmospheric pressure sensor 23 detects the atmospheric pressure around the terminal device 10, for example.

Since the terminal device 10 is equipped with the above-mentioned acceleration sensor 21, gyro sensor 22, atmospheric pressure sensor 23, etc., it is possible to implement technologies such as pedestrian autonomous navigation (PDR) using these sensors 21 to 23, etc. It becomes possible to measure the position of the terminal device 10 using the . This makes it possible to obtain indoor position information that is difficult to obtain using positioning systems such as GPS.

For example, a pedometer using the acceleration sensor 21 can calculate the number of steps, walking speed, and distance walked. Furthermore, by using the gyro sensor 22, it is possible to know the direction of travel of the user U, the direction of the line of sight, and the inclination of the user's body. Further, from the atmospheric pressure detected by the atmospheric pressure sensor 23, it is also possible to know the altitude and the number of floors where the terminal device 10 of the user U is located.

The temperature sensor 24 detects, for example, the temperature around the terminal device 10. The sound sensor 25 detects, for example, sounds around the terminal device 10. The optical sensor 26 detects the illuminance around the terminal device 10 . The magnetic sensor 27 detects, for example, the earth's magnetism around the terminal device 10. The image sensor 28 captures an image of the surroundings of the terminal device 10.

The above-mentioned atmospheric pressure sensor 23, temperature sensor 24, sound sensor 25, optical sensor 26, and image sensor 28 each detect atmospheric pressure, temperature, sound, and illuminance, and capture images of the surroundings, so that the terminal device 10 It is possible to detect the surrounding environment and situation. Furthermore, it is possible to improve the accuracy of the location information of the terminal device 10 based on the environment and situation around the terminal device 10.

(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, etc., and various circuits. Further, the control unit 30 may be configured with hardware such as an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The control section 30 includes a transmitting section 31, a receiving section 32, and a processing section 33.

(Transmission unit 31)
The transmitting unit 31 receives, for example, various information input by the user U using the input unit 13, various information detected by the sensors 21 to 28 mounted on or connected to the terminal device 10, and information measured by the positioning unit 14. The location information of the terminal device 10 and the like can be transmitted to the information providing device 100 via the communication unit 11.

(Receiving unit 32)
The receiving unit 32 can receive various information provided from the information providing device 100 and requests for various information from the information providing device 100 via the communication unit 11 .

(Processing unit 33)
The processing unit 33 controls the entire terminal device 10, including the display unit 12 and the like. For example, the processing unit 33 can output various types of information transmitted by the transmitting unit 31 and various types of information received by the receiving unit 32 from the information providing device 100 to the display unit 12 for display.

(Storage unit 40)
The storage unit 40 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), or an optical disk. Ru. The storage unit 40 stores various programs, various data, and the like.

[4. Configuration example of information providing device]
Next, the configuration of the information providing apparatus 100 according to the embodiment will be described using FIG. 5. FIG. 5 is a diagram illustrating a configuration example of the information providing apparatus 100 according to the embodiment. As shown in FIG. 5, the information providing device 100 includes a communication section 110, a storage section 120, and a control section 130.

(Communication Department 110)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card). Further, the communication unit 110 is connected to the network N (see FIG. 3) by wire or wirelessly.

(Storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as an HDD, an SSD, or an optical disk. As shown in FIG. 5, the storage unit 120 includes a user information database 121, a history information database 122, and a relevance information database 123.

(User information database 121)
The user information database 121 stores user information regarding the user U. For example, the user information database 121 stores various information such as user U's attributes. FIG. 6 is a diagram showing an example of the user information database 121. In the example shown in FIG. 6, the user information database 121 includes items such as "user ID (identifier),""age,""gender,""home,""worklocation," and "interest."

“User ID” indicates identification information for identifying user U. Note that the "user ID" may be user U's contact information (telephone number, email address, etc.), or may be identification information for identifying user U's terminal device 10.

Moreover, "age" indicates the age of the user U identified by the user ID. Note that the "age" may be information indicating the specific age of the user U (for example, 35 years old, etc.), or may be information indicating the age of the user U (for example, 30s, etc.) . Alternatively, the "age" may be information indicating the date of birth of the user U, or may be information indicating the generation of the user U (for example, born in the 1980s). Furthermore, “gender” indicates the gender of the user U identified by the user ID.

Moreover, "home" indicates the location information of the home of the user U identified by the user ID. In the example shown in FIG. 6, "home" is illustrated as an abstract code such as "LC11," but it may also be latitude and longitude information. Furthermore, for example, "home" may be a region name or address.

Moreover, "work place" indicates the location information of the work place (school in the case of a student) of the user U identified by the user ID. In the example shown in FIG. 6, the "work location" is illustrated as an abstract code such as "LC12," but it may also be latitude and longitude information. Further, for example, the "work location" may be a region name or address.

Moreover, "interest" indicates the interest of the user U identified by the user ID. That is, "interest" indicates an object in which the user U identified by the user ID has a high interest. For example, "interest" may be a search query (keyword) that the user U inputs into a search engine. In the example shown in FIG. 6, one "interest" is shown for each user U, but there may be a plurality of "interests".

For example, in the example shown in FIG. 6, the age of the user U identified by the user ID "U1" is "20s", and the gender is "male". Further, for example, the user U identified by the user ID "U1" indicates that his home is "LC11". Further, for example, the user U identified by the user ID "U1" indicates that the work location is "LC12". Further, for example, the user U identified by the user ID "U1" indicates that he is interested in "sports."

Here, in the example shown in FIG. 6, abstract values such as "U1", "LC11", and "LC12" are used for illustration, but "U1", "LC11", and "LC12" have specific values. It is assumed that information such as character strings and numerical values is stored. Below, abstract values may be illustrated in diagrams related to other information as well.

Note that the user information database 121 is not limited to the above, and may store various information depending on the purpose. For example, the user information database 121 may store various information regarding the terminal device 10 of the user U. In addition, the user information database 121 includes information such as demographic (demographic attributes), psychographic (psychological attributes), geographic (geographical attributes), behavioral (behavioral attributes), etc. of user U. Information regarding attributes may also be stored. For example, the user information database 121 includes name, family composition, place of birth (locality), occupation, position, income, qualifications, type of residence (single-family house, condominium, etc.), presence or absence of a car, commuting/commuting time, commuting/commuting, etc. Memorizes information such as routes, commuter pass sections (stations, lines, etc.), frequently used stations (other than the stations closest to your home or work), lessons (location, time zone, etc.), hobbies, interests, lifestyle, etc. You can.

(History information database 122)
The history information database 122 stores various information related to history information (log data) indicating the actions of the user U. FIG. 7 is a diagram showing an example of the history information database 122. In the example shown in FIG. 7, the history information database 122 has items such as "user ID", "location history", "search history", "browsing history", "purchase history", and "posting history".

“User ID” indicates identification information for identifying user U. Further, “position history” indicates a position history that is a history of the user U's position and movement. Further, “search history” indicates a search history that is a history of search queries input by the user U. In addition, “browsing history” indicates a browsing history that is a history of contents that the user U has viewed. Further, “purchase history” indicates a purchase history that is a history of purchases by user U. Moreover, "posting history" indicates a posting history that is a history of postings by user U. Note that the "posting history" may include questions regarding user U's belongings.

For example, in the example shown in FIG. 7, the user U identified by the user ID "U1" moves as per "location history #1", searches as per "search history #1", and searches as per "view history #1". This indicates that the content was viewed according to "history #1", a predetermined product, etc. was purchased at a predetermined store etc. according to "purchase history #1", and the content was posted according to "posting history".

Here, in the example shown in FIG. 7, abstracts such as "U1", "location history #1", "search history #1", "browsing history #1", "purchase history #1" and "posting history #1" are used. Although the figures are shown using typical values, "U1", "location history #1", "search history #1", "browsing history #1", "purchase history #1" and "posting history #1" are It is assumed that information such as specific character strings and numerical values is stored.

Note that the history information database 122 is not limited to the above, and may store various information depending on the purpose. For example, the history information database 122 may store the user U's usage history of a predetermined service. Further, the history information database 122 may store the user U's visit history to a physical store, visit history to a facility, or the like. Further, the history information database 122 may store the payment history of user U's payment using the terminal device 10 (electronic payment).

(Relevance information database 123)
The relationship information database 123 stores various information regarding relationships between entities. FIG. 8 is a diagram showing an example of the relevance information database 123. In the example shown in FIG. 8, the relevance information database 123 has items such as "word", "candidate", and "probability".

"Word" indicates a word included in the query. “Candidate” indicates a candidate word adjacent to the word. "Probability" indicates the probability of occurrence of an entity for a word sequence of the word and the candidate. The information providing device 100 calculates the score of the entity series based on this probability and the inner product of the entities.

For example, in the example shown in FIG. 8, the probability of occurrence of the entity "probability #A1" for the word series of the word "A" and the adjacent word candidate "A1" is listed.

Here, the example shown in FIG. 8 is illustrated using abstract values such as "A", "A1", and "probability #A1", but "A", "A1", and "probability #A1" are It is assumed that information such as specific character strings and numerical values is stored.

Note that the relevance information database 123 is not limited to the above, and may store various information depending on the purpose. For example, the relevance information database 123 may store information regarding a lattice structure of entity candidates of a word series. Further, the relevance information database 123 may store information regarding a model that models a query as a sentence. Further, the relevance information database 123 may store calculated inner products, scores, and the like.

(Control unit 130)
Returning to FIG. 5, the explanation will be continued. The control unit 130 is a controller, and uses, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field Programmable Gate Array) to control the information providing apparatus 100. This is achieved by executing various programs (corresponding to an example of an information processing program) stored in an internal storage device using a storage area such as a RAM as a work area. In the example shown in FIG. 5, the control unit 130 includes an acquisition unit 131, a calculation unit 132, and a provision unit 133.

(Acquisition unit 131)
The acquisition unit 131 acquires a search query input by the user U. For example, when the user U inputs a search query into a search engine or the like and performs 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 window of a search engine, site, or application. That is, the acquisition unit 131 functions as a reception unit that receives input of a query.

The acquisition unit 131 also acquires user information regarding the user U via the communication unit 110. For example, the acquisition unit 131 acquires identification information indicating the user U (user ID, etc.), location information of the user U, attribute information of the user U, etc. from the terminal device 10 of the user U. Further, the acquisition unit 131 may acquire identification information indicating the user U, attribute information of the user U, etc. at the time of user U's user registration. The acquisition unit 131 then registers the user information in the user information database 121 of the storage unit 120.

Further, 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 the user ID or the like. The acquisition unit 131 then registers various types of history information in the history information database 122 of the storage unit 120.

(Calculation unit 132)
The calculation unit 132 constructs a lattice structure using the candidate entities of the word sequence of the query, and obtains the most likely entity sequence. For example, the calculation unit 132 enumerates candidate entities along with their probabilities for the word series, calculates the relationship between the candidate entities using the inner product of the virtual document, takes the logarithmic sum of the probability and the inner product, and arranges the virtual nodes before and after the lattice. , and find the path with the maximum score between virtual nodes to obtain the maximum likelihood entity sequence.

At this time, the calculation unit 132 applies Japanese morphological analysis using CRF to the word sequence of the query. Furthermore, the calculation unit 132 estimates a word that is not associated with an entity as a virtual document consisting of the word itself. Further, the calculation unit 132 evaluates the overall consistency by treating entities and word sequences in a unified manner.

Further, the calculation unit 132 calculates a vector representation of the word sequence and evaluates the consistency. Further, the calculation unit 132 identifies an entity from a word sequence of arbitrary length. Furthermore, the calculation unit 132 models the query as a sentence, taking into account the relationship between adjacent words.

Then, the calculation unit 132 associates the calculated probability with the candidate entity for the word series and registers it in the relevance information database 123 of the storage unit 120. Further, the calculation unit 132 temporarily stores the calculated probability, inner product, score, etc. in a memory, cache, etc. for processing.

(Providing unit 133)
The providing unit 133 provides, via the communication unit 110, the search result corresponding to the most likely entity series to the terminal device 10 of the user U who has input the search query.

[5. Processing procedure]
Next, a processing procedure by the information providing apparatus 100 according to the embodiment will be described using FIG. 9. FIG. 9 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 information providing device 100.

As shown in FIG. 9, the acquisition unit 131 of the information providing device 100 acquires the input search query from the terminal device 10 of the user U via the communication unit 110 (step S101). That is, the acquisition unit 131 receives a search query input by the user U as a reception unit.

Subsequently, the calculation unit 132 of the information providing device 100 enumerates candidate entities along with probabilities for the word series of the query, and constructs a lattice structure using the candidate entities of the word series of the query (step S102).

Subsequently, the calculation unit 132 of the information providing apparatus 100 calculates the relationship between the candidate entities using the inner product of the virtual document (step S103).

Next, the calculation unit 132 of the information providing device 100 takes the logarithmic sum of the probability (<=1) and the inner product (<=1), and obtains the score of the route expressed by the accumulation of the probability and the inner product (step S104 ).

Subsequently, the calculation unit 132 of the information providing apparatus 100 installs virtual nodes before and after the lattice (step S105). For example, the calculation unit 132 places virtual nodes "start" and "end" before and after the lattice.

Subsequently, the calculation unit 132 of the information providing device 100 finds a route with the maximum score among the virtual nodes, and obtains the most likely entity sequence (step S106). For example, the calculation unit 132 finds a route with the maximum score between the virtual nodes "start" and "end".

At this time, the calculation unit 132 applies Japanese morphological analysis using CRF to the word sequence of the query. Furthermore, the calculation unit 132 estimates a word that is not associated with an entity as a virtual document consisting of the word itself. Further, the calculation unit 132 evaluates the overall consistency by treating entities and word sequences in a unified manner.

Thereby, the calculation unit 132 calculates a vector representation of the word sequence and evaluates the consistency. Further, the calculation unit 132 identifies an entity from a word sequence of arbitrary length. Furthermore, the calculation unit 132 models the query as a sentence, taking into account the relationship between adjacent words.

Subsequently, the providing unit 133 of the information providing device 100 provides the terminal device 10 of the user U who has input the search query via the communication unit 110 with a search result corresponding to the most likely entity series (step S10 ).

[6. Modified example]
The terminal device 10 and the information providing device 100 described above may be implemented in various different forms other than the above embodiments. Therefore, a modification of the embodiment will be described below.

In the above embodiment, part or all of the processing executed by the information providing device 100 may actually be executed by the terminal device 10. For example, the process may be completed stand-alone (by the terminal device 10 alone). In this case, it is assumed that the terminal device 10 has the functions of the information providing device 100 in the above embodiment. Furthermore, in the embodiment described above, since the terminal device 10 cooperates with the information providing device 100, it appears to the user U that the terminal device 10 also executes the processing of the information providing device 100. That is, from another perspective, it can be said that the terminal device 10 includes the information providing device 100.

Furthermore, in the above embodiment, the information providing apparatus 100 may consider attribute information, history information, etc. of the user U who inputs the query when calculating the most likely entity sequence. For example, the information providing device 100 may weight the occurrence probability of an entity for a word series, the score of a route, etc., depending on the attribute information, history information, etc. of the user U who inputs the query.

[7. effect〕
As described above, the information processing device (terminal device 10 and information providing device 100) according to the present application constructs a lattice structure using the reception unit (acquisition unit 131) that receives query input and candidate entities of the word series of the query. The present invention is characterized in that it includes a calculation unit 132 that calculates the most likely entity sequence, and a providing unit 133 that provides search results according to the most likely entity sequence.

The calculation unit 132 enumerates candidate entities with their probabilities for the word series, calculates the relationship between the candidate entities using the inner product of the virtual document, calculates the logarithmic sum of the probability and the inner product, and places virtual nodes before and after the lattice. Then, find the path with the maximum score between virtual nodes to obtain the maximum likelihood entity sequence.

The calculation unit 132 estimates a word to which no entity is associated as a virtual document consisting of the word itself.

The calculation unit 132 evaluates the overall consistency by treating entities and word sequences in a unified manner.

The calculation unit 132 applies Japanese morphological analysis using CRF to the word sequence of the query.

The calculation unit 132 calculates a vector representation of the word sequence and evaluates the consistency.

The calculation unit 132 identifies entities from a word sequence of arbitrary length.

The calculation unit 132 models the query as a sentence, taking into account the relationship between adjacent words.

By using any one or a combination of the above-described processes, the information processing apparatus according to the present application can realize robust entity linking using context consistency.

[8. Hardware configuration]
Further, the terminal device 10 and the information providing device 100 according to the embodiments described above are realized, for example, by a computer 1000 having a configuration as shown in FIG. The information providing apparatus 100 will be described below as an example. FIG. 10 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 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 to a bus. 1090.

The arithmetic unit 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 performs 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.

The primary storage device 1040 is a memory device such as a RAM (Random Access Memory) that temporarily stores data used by the calculation device 1030 for various calculations. Further, the secondary storage device 1050 is a storage device in which data used by the calculation device 1030 for various calculations and various databases are registered, and includes a ROM (Read Only Memory), an HDD (Hard Disk Drive), and an SSD (Solid Disk Drive). This is realized using flash memory, etc. The secondary storage device 1050 may be a built-in storage or an external storage. Further, 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. Further, the secondary storage device 1050 may be a cloud storage (online storage), a NAS (Network Attached Storage), a file server, or the like.

The output I/F 1060 is an interface for transmitting information to be output to the output device 1010 that outputs various information such as a display, a projector, and a printer. (Digital Visual Interface) and HDMI (registered trademark) (High Definition Multimedia Interface). Further, the input I/F 1070 is an interface for receiving information from various input devices 1020 such as a mouse, keyboard, keypad, button, scanner, etc., and is realized by, for example, a USB or the like.

Further, 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, output device 1010 and input device 1020 may be wireless devices.

Moreover, the output device 1010 and the input device 1020 may be integrated like a touch panel. In this case, the output I/F 1060 and the input I/F 1070 may also be integrated as an input/output I/F.

Note that the input device 1020 is, 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), or a tape. It may be a device that reads information from a medium, a magnetic recording medium, a semiconductor memory, or the like.

Network I/F 1080 receives data from other devices via network N and sends it to computing device 1030, and also sends data generated by computing device 1030 to other devices via network N.

Arithmetic device 1030 controls output device 1010 and input device 1020 via output I/F 1060 and input I/F 1070. For example, the arithmetic device 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.

For example, when the computer 1000 functions as the information providing device 100, the arithmetic unit 1030 of the computer 1000 realizes the functions of the control unit 130 by executing a program loaded onto the primary storage device 1040. Further, the arithmetic device 1030 of the computer 1000 may load a program obtained from another device via the network I/F 1080 onto the primary storage device 1040, and execute the loaded program. Furthermore, the arithmetic unit 1030 of the computer 1000 may cooperate with other devices via the network I/F 1080, and may call and use program functions, data, etc. from other programs of other devices.

[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. Furthermore, the above-mentioned components include those that can be easily assumed by those skilled in the art, those that are substantially the same, and those that are in a so-called equivalent range. Furthermore, the aforementioned components can be combined as appropriate. Furthermore, various omissions, substitutions, or modifications of the constituent elements can be made without departing from the gist of the embodiments described above.

Further, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or the processes described as being performed manually can be performed manually. All or part of this can also be performed automatically using known methods. In addition, information including the processing procedures, specific names, and various data and parameters shown in the above documents and drawings may be changed arbitrarily, unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

Furthermore, each component of each device shown in the drawings is functionally conceptual, and does not necessarily need to be physically configured as shown in the drawings. In other words, the specific form of distributing and integrating each device is not limited to what is shown in the diagram, and all or part of the devices can be functionally or physically distributed or integrated in arbitrary units depending on various loads and usage conditions. Can be integrated and configured.

For example, the information providing apparatus 100 described above may be realized by a plurality of server computers, and depending on the function, it may be realized by calling an external platform etc. using an API (Application Programming Interface), network computing, etc. can be changed flexibly.

Furthermore, the above-described embodiments and modifications can be combined as appropriate within a range that does not conflict with the processing contents.

Further, the above-mentioned "section, module, unit" can be read as "means", "circuit", etc. For example, the acquisition unit can be read as an acquisition means or an acquisition circuit.

1 Information processing system 10 Terminal device 100 Information providing device 110 Communication unit 120 Storage unit 121 User information database 122 History information database 123 Relevance information database 130 Control unit 131 Acquisition unit 132 Calculation unit 133 Provision unit

Claims (9)

a reception section that accepts query input;
a calculation unit that constructs a lattice structure using candidate entities of the word series of the query and calculates a maximum likelihood entity series;
a providing unit that provides search results according to the most likely entity series;
Equipped with
The calculation unit enumerates candidate entities with probabilities for the word series, calculates the relationship between the candidate entities by an inner product of the virtual document, calculates the logarithmic sum of the probability and the inner product, and divides the virtual nodes before and after the lattice. An information processing device installed in The information processing device according to claim 1, wherein the calculation unit estimates a word that is not associated with an entity as a virtual document consisting of the word itself. The information processing device according to claim 1 or 2 , wherein the calculation unit applies Japanese morphological analysis using CRF to a word sequence of a query. When there is an entity associated with a word series in the knowledge base, the calculation unit calculates a topic vector of a virtual document consisting of all click logs associated with the entity and evaluates context consistency.
The information processing device according to any one of claims 1 to 3, characterized in that: 5. The calculation unit evaluates context consistency by calculating a topic vector of a virtual document consisting of a single word when there is no entity to which a word sequence is linked in the knowledge base. The information processing device according to any one of them. The calculation unit identifies an entity from a word sequence of arbitrary length, and is capable of identifying not only words separated by spaces but also word strings. The first pattern is divided into three parts, "A", "B", and "C", with "C" separated by spaces; the second pattern is divided into two parts, "A B" and "C", and one part, "A". ” and “B C”, the third pattern is divided into one and two, and the fourth pattern is fixed to one, “A B C”. Calculate each of the four patterns and return the one with the highest score.
The information processing device according to any one of claims 1 to 5. 2. The calculation unit, when modeling an entity as an LDA document generated from a clicked word, considers the relationship between adjacent words and models the query as a sentence. The information processing device according to any one of -6. An information processing method executed by an information processing device, the method comprising:
a reception process for accepting input of a query;
a calculation step of constructing a lattice structure using the candidate entities of the word series of the query and obtaining a maximum likelihood entity series;
a providing step of providing search results according to the most likely entity series;
including;
In the calculation step, candidate entities are listed with their probabilities for the word series, the relationship between the candidate entities is calculated by the inner product of the virtual document, the logarithmic sum of the probability and the inner product is taken, and the virtual nodes are arranged before and after the lattice. An information processing method characterized in that the route with the maximum score is found between virtual nodes, and the entity sequence with the maximum likelihood is obtained. A reception procedure for accepting query input;
a calculation procedure for constructing a lattice structure using candidate entities of the word series of the query and obtaining a maximum likelihood entity series;
a providing step of providing search results according to the most likely entity series;
An information processing program for causing a computer to execute
In the calculation procedure, candidate entities are listed with their probabilities for the word series, the relationship between the candidate entities is calculated by the inner product of the virtual document, the logarithmic sum of the probability and the inner product is taken, and the virtual nodes are arranged before and after the lattice. An information processing program that is installed in a virtual node, finds a route with the maximum score between virtual nodes, and obtains a maximum likelihood entity sequence.

Images