WO2016189905A1 - Dispositif de traitement d'informations, procédé de traitement d'informations et programme - Google Patents

Dispositif de traitement d'informations, procédé de traitement d'informations et programme Download PDF

Info

Publication number
WO2016189905A1
WO2016189905A1 PCT/JP2016/054961 JP2016054961W WO2016189905A1 WO 2016189905 A1 WO2016189905 A1 WO 2016189905A1 JP 2016054961 W JP2016054961 W JP 2016054961W WO 2016189905 A1 WO2016189905 A1 WO 2016189905A1
Authority
WO
WIPO (PCT)
Prior art keywords
models
model
user
information
information processing
Prior art date
Application number
PCT/JP2016/054961
Other languages
English (en)
Japanese (ja)
Inventor
淳史 野田
勝吉 金本
慎吾 高松
拓也 藤田
Original Assignee
ソニー株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ソニー株式会社 filed Critical ソニー株式会社
Publication of WO2016189905A1 publication Critical patent/WO2016189905A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor

Definitions

  • the present disclosure relates to an information processing apparatus, an information processing method, and a program.
  • Patent Literature 1 discloses a technique for increasing the possibility that recommended content or the like is accepted by a user.
  • Patent Document 1 the technology proposed in the above-mentioned Patent Document 1 is still short since it was developed, and it is said that the technology for providing appropriate information to users in various aspects has been sufficiently proposed. It's hard. For example, a technique for improving the accuracy of information provided in response to feedback from a user is one that has not been sufficiently proposed.
  • the present disclosure proposes a new and improved information processing apparatus, information processing method, and program capable of providing more appropriate information to the user.
  • a providing unit that provides a user with information generated using one model selected from a plurality of models to be selected by a model selection algorithm, and a user's response to the information provided by the providing unit
  • An information processing apparatus includes a generation unit that generates information to be provided to a user in response to feedback.
  • an information processing method including generating information to be provided to a user by a processor is provided.
  • a computer is provided by a providing unit that provides a user with information generated using one model selected from a plurality of models to be selected by a model selection algorithm, and the providing unit provides the computer.
  • a program for functioning as a generation unit that generates information to be provided to the user in response to user feedback on the received information is provided.
  • elements having substantially the same functional configuration may be distinguished by adding different alphabets after the same reference numerals.
  • a plurality of elements having substantially the same functional configuration are distinguished as the terminal devices 100A and 100B as necessary.
  • only the same reference numerals are given.
  • the terminal device 100 when it is not necessary to distinguish between the terminal devices 100A and 100B, they are simply referred to as the terminal device 100.
  • FIG. 1 is an explanatory diagram for explaining an overview of the information providing technology according to the present embodiment.
  • a plurality of models 10 that is, models 10 ⁇ / b> A and 10 ⁇ / b> B are assigned to the model selection algorithm 20.
  • the model 10 is a model for generating information to be provided to the user.
  • the model 10 may be a model of collaborative filtering (CF: Collaborative Filtering) or content-based filtering (CBF: content-based filtering).
  • the information provided to the user is also referred to as an item below.
  • the model 10 is information in which preference information for each of a plurality of items is accumulated for each of a plurality of users.
  • the model 10 includes an evaluation value (hereinafter also referred to as a content vector) of each feature of each of a plurality of items, and user preference information (hereinafter referred to as a user). (Also referred to as profile vector).
  • the model 10 may be a bandit algorithm (BA: Bandit Algorithms) or an A / B testing (A / B testing).
  • the model selection algorithm 20 is an algorithm for selecting any one model 10 by selecting a plurality of assigned models 10 as selection targets.
  • the model selection algorithm 20 may be a BA or A / B test or the like.
  • BA is a score indicating the priority of each of the plurality of selection objects based on the results obtained when a plurality of selection objects are selected and selected (the actual value assigned to each of the selection objects). It is an algorithm that maximizes the gain obtained by updating (numerical value).
  • the score indicating the priority may be, for example, an expected value, an expected value + search term, or a real value calculated by any other method.
  • the model 10 with the better score is preferentially selected. In this specification, it is assumed that the model 10 is better as the score is higher.
  • the selection object in BA is also referred to as an arm.
  • the BA may assign an exploit score and an exploration score to each arm.
  • the utilization score takes a higher value as the expectation that a good result can be obtained at the present time is larger.
  • the search score takes a high value when the reliability of the utilization score for the arm is low.
  • BA selects the arm having the largest sum of the utilization score and the search score. Thereby, in BA, utilization and search can be balanced.
  • THOMPSON SAMPLING the posterior distribution of the utilization score is used instead of the search score, thereby providing a search effect.
  • the A / B test is an algorithm similar to BA.
  • the model selection algorithm 20 will be described as being BA.
  • an item generated by the arm (model 10) selected by the model selection algorithm 20 is provided.
  • the user provides feedback on the provided item.
  • feedback for example, when the item is news, whether or not the provided news has been read, the ratio of reading the whole, and the like can be considered.
  • the feedback may be whether or not the user has performed a positive action (for example, posting a comment or pressing a like button) on the provided item.
  • learning is performed based on such feedback. Specifically, the score of the selected arm is updated, and the parameters of the selected model 10 and the unselected model 10 are learned.
  • the learning process will be described in detail with reference to FIG.
  • FIG. 2 is an explanatory diagram for explaining a learning process in the information providing technique according to the present embodiment.
  • the models 10A and 10B shown in FIG. 2 are CF models with different parameters (scores for each item for each user).
  • the score of the model 10 indicates, for example, the degree of expectation that an item provided using each model 10 receives positive feedback from the user. Note that this score can also be taken as a parameter of the model 10.
  • the model 10A is selected by the model selection algorithm 20 and the item B is provided to the user A.
  • the score for the item B of the user A in the model 10A is updated based on the feedback of the user A for the provided item B, and the score of the model 10A is updated.
  • the score for the item B of the user A in the model 10B is updated.
  • the information providing technique according to the present embodiment it is possible to perform learning as if the model 10 that was not selected is selected. As a result, learning of the plurality of models 10 to be selected is efficiently performed, and which model 10 should be used to provide an item is efficiently determined.
  • the information providing technology according to the present embodiment can provide more appropriate information to the user.
  • the information providing technique according to the present embodiment does not depend on a model prepared in advance and can avoid the cold start problem.
  • FIG. 3 is a block diagram illustrating a configuration example of the information processing system 1 according to the present embodiment.
  • the information processing system 1 according to the present embodiment includes one or more terminal devices 100 (that is, terminal devices 100 ⁇ / b> A and 100 ⁇ / b> B) and a server 200.
  • the terminal device 100 is a device that collects information about the user and outputs information to the user. For example, the terminal device 100 outputs the item generated by the server 200 to the user, and transmits the collected information about the user to the server 200.
  • the terminal device 100 is a device such as a smartphone, a mobile phone, a PC, or a notebook PC.
  • the terminal device 100 may be an environment-installed device.
  • the environment-installed device refers to a device that is fixedly or semi-fixedly provided in real space, such as a digital signage and a surveillance camera.
  • the server 200 is a device that generates items and provides them to the user. As illustrated in FIG. 2, the server 200 includes a communication unit 210, a storage unit 220, and a processing unit 230.
  • the communication unit 210 is a wired or wireless communication module for transmitting and receiving data between the server 200 and another device.
  • the communication unit 210 communicates with the terminal device 100 via a LAN (Local Area Network) or a telephone line.
  • LAN Local Area Network
  • the storage unit 220 temporarily or permanently stores a program for operating the server 200 and various data.
  • the storage unit 220 stores information on a model assigned to a model selection algorithm for each user.
  • the processing unit 230 provides various functions of the server 200. As illustrated in FIG. 3, the processing unit 230 includes an allocation unit 231, a generation unit 232, a provision unit 233, and a learning unit 234. Note that the processing unit 230 may further include other components other than these components. That is, the processing unit 230 can perform operations other than the operations of these components.
  • the assigning unit 231 has a function of assigning a plurality of models to be selected to the model selection algorithm.
  • the generating unit 232 has a function of generating an item provided by the providing unit 233 in response to a user feedback on the item provided by the providing unit 233. Specifically, the generation unit 232 selects a model whose parameters have been updated by the learning unit 234 according to feedback using a model selection algorithm, and generates an item using the selected model. Therefore, the item provided to the user is changed according to the feedback from the user.
  • the providing unit 233 has a function of providing the user with the item generated by the generating unit 232. For example, the providing unit 233 notifies the terminal device 100 of the item via the communication unit 210.
  • the learning unit 234 has a function of learning each of a plurality of models assigned to the model selection algorithm. For example, the learning unit 234 learns each parameter of a plurality of models to be selected in accordance with user feedback on the information provided by the providing unit 233.
  • the outline of learning is as described above with reference to FIG.
  • the server 200 (for example, the assigning unit 231) can assign various models as a plurality of models to be selected to the model selection algorithm.
  • the plurality of models to be selected may include different types of models.
  • a plurality of models to be selected include contextual BA (for example, LinUCB method), non-contextual BA (for example, Thompson sampling method or Epsilon-Greedy method), It can include different types of models, such as CF and CBF.
  • the server 200 can generate and provide items based on various criteria.
  • a plurality of models to be selected may include models having the same use but different types.
  • uses include recommendation and prediction.
  • Examples of different types for recommended use include BA, CF, and CBF models.
  • Examples of different models for prediction use include a moving average method and exponential smoothing model for sales prediction.
  • the server 200 can generate and provide items based on various criteria in order to satisfy an arbitrary usage.
  • the plurality of models to be selected may include models with different uses and types.
  • the plurality of models to be selected may include a BA model for recommended use, a moving average method model for prediction use, and the like.
  • the server 200 can generate and provide items based on various criteria.
  • the plurality of models to be selected may include models of the same type but different parameters.
  • the server 200 can search for an optimal parameter based on the scores of the same type of model.
  • the plurality of models to be selected may include a model selection algorithm that selects a lower model as a selection target. That is, the selection target of the model selection algorithm may have a hierarchical structure.
  • FIG. 4 is an explanatory diagram for explaining an example of assignment of models to be selected according to the present embodiment. As shown in FIG. 4, models 10A and 10B are assigned to the model selection algorithm 20A, and models 10C and 10D are assigned to the model selection algorithm 20B. The model selection algorithms 20A and 20B are assigned to the model selection algorithm 20C. When the selection target of the model selection algorithm has such a hierarchical structure, the server 200 can perform more flexible learning and item generation and provision.
  • the score may be calculated only for the lowest layer model (for example, the models 10A, 10B, 10C, and 10D), or the model selection algorithm for the intermediate layer (for example, The model selection algorithms 20A and 20B) may also be calculated.
  • the score of the lowermost model represents how good each model of the lowermost layer is.
  • the score of the model selection algorithm in the intermediate layer represents how good each model selection algorithm in the intermediate layer is.
  • the plurality of models to be selected may include models with different learning progresses.
  • the server 200 can avoid the cold start problem by assigning a learned model assigned to an unlearned user with respect to a user who has already learned.
  • the plurality of models to be selected may include models that are included in common among different users. That is, the server 200 may share the same model among a plurality of users.
  • FIG. 5 is an explanatory diagram for explaining an example of assignment of models to be selected according to the present embodiment.
  • the model selection algorithm 20A is a model selection algorithm for the user A, and models 10A and 10B are assigned thereto.
  • the model selection algorithm 20B is a model selection algorithm for the user B, and models 10B and 10C are assigned thereto.
  • the model 10 ⁇ / b> B is included in common between the user A and the user B.
  • the server 200 can efficiently learn the model 10B based on feedback from both the user A and the user B.
  • the server 200 can use the model 10B to select an appropriate item with a certain accuracy for the user A. It becomes possible to provide. Moreover, if learning progresses about the model 10A, the server 200 provides the user A with a more appropriate item than using the model 10B by using the model 10A learned only based on feedback from the user A. Is possible.
  • the plurality of models to be selected may include a model that considers the user context and a model that does not consider the user context.
  • the plurality of models to be selected may include models of contextual BA and non-contextual BA.
  • the server 200 can provide an appropriate item regardless of whether the user has a different or different preference depending on the context.
  • the context in this specification refers to the user's actions (walking, sitting, running, etc.), the user's location (home, work, on the train), time zone, day of the week, and who Or information indicating the environment of the user himself or his / her surroundings such as biological information.
  • the plurality of models to be selected include models corresponding to user attribute information.
  • a model corresponding to attribute information such as the user's sex, age, and work can be assigned.
  • the model corresponding to the attribute information may be a model that has already been learned with respect to another user having the same attribute information, for example.
  • the server 200 can provide an appropriate item according to the user's attribute information to the user.
  • the server 200 may exclude the model from the selection target or add a new model according to the scores of the plurality of models to be selected in the model selection algorithm. For example, the server 200 excludes models whose number of selections is greater than or equal to a threshold and whose score is less than or equal to the threshold from the selection target. This prevents an inappropriate item from being provided to the user. Further, from experience, learning efficiency can be improved as the number of models to be selected decreases. Instead, the server 200 may add a new model to the selection target, or may add a new model to the selection target without excluding it.
  • various models can be included in the selection target. Furthermore, models that do not fit the user in the course of learning are excluded and new models are added. For example, a model that fits the user remains in the selection target. Thereby, in this information provision technique, it becomes possible to provide an appropriate item to a user without depending on the model prepared in advance.
  • the server 200 learns a plurality of models to be selected by the model selection algorithm.
  • the server 200 learns parameters of each of a plurality of models to be selected in accordance with user feedback on the provided information. For example, in the example described above with reference to FIG. 2, the server 200 uses the scores of all the CF models (models 10A and 10B) assigned to the model selection algorithm 20 and the scores of the selected models as the user. Update according to feedback. It is possible to learn a plurality of models to be selected as a whole when the selection targets are all of the same type as shown in FIG. 2, and as shown in FIG. It is possible even when different types of models are included.
  • FIG. 6 is an explanatory diagram for explaining a learning process in the information providing technique according to the present embodiment.
  • a model 10A shown in FIG. 6 is a CF model.
  • the model 10B is a CBF model including a user profile vector indicating a user's preference with respect to an item feature, and a content vector indicating a feature for each item. In FIG. 6, description regarding the score is omitted.
  • the server 200 can learn a plurality of models to be selected as a whole.
  • CF and CBF are mixed in the selection target
  • learning is possible in the same manner even in a combination of other arbitrary types of models, and three or more types of models are mixed. Can learn as well.
  • the server 200 executes learning for each user as a learning policy. Thereby, the server 200 can generate and provide appropriate information for each user.
  • the selection target may include a model included in common among different users.
  • the server 200 may reflect the learning result based on the feedback of one user also about other users about the model included in common between different users.
  • the server 200 can efficiently learn about the commonly included models. For example, in the example illustrated in FIG. 5, the server 200 learns the model 10B based on feedback from both the user A and the user B.
  • the server 200 may reflect the learning result for other users only with respect to feedback from some users regarding models included in common among different users. For example, in the example illustrated in FIG. 5, the server 200 may perform learning of the model 10B based only on feedback from the user B. In this case, the server 200 can provide an appropriate item with a certain accuracy to the user A, and originally avoids the influence of the other users on the model 10B for the user B. can do.
  • the server 200 (for example, the generation unit 232 and the providing unit 233) generates items according to user feedback and provides them to the user. That is, the item provided to the user is changed according to the feedback from the user.
  • the server 200 selects a model having a high score from a plurality of models to be selected by a model selection algorithm, and generates an item.
  • the server 200 may provide information indicating the reason for generating the item together with the item or instead of the item. Specifically, the server 200 may provide information indicating characteristics of a model used for generating an item among a plurality of models to be selected. Examples of model characteristics include model type, application, parameter bias, and learning progress. By providing the generation reason, it is considered that the user can easily accept the provided item.
  • the server 200 when an item generated using a model having a high score is provided, the server 200 provides information indicating a feature of a model having a high score among a plurality of models to be selected as information indicating a generation reason. May be.
  • the server 200 By knowing the characteristics of a model with a high score, that is, a model with a strong tendency to give positive feedback, the user can know his / her own tendency.
  • the high score may be taken as an arbitrary meaning such as the highest score among a plurality of models to be selected, or the score exceeding a predetermined threshold.
  • the server 200 may not provide information indicating the generation reason. This is because a model with a low score is not a model fitted to the user, and even if information indicating the feature is provided, the user may be confused.
  • a UI (User Interface) example in the case where a generation reason is provided together with an item will be described with reference to FIG. 7 and FIG. Note that the UI example described in this section is displayed on the terminal device 100, for example.
  • FIG. 7 is a diagram for explaining an example of a UI according to the present embodiment.
  • UI example 41 shows a display example of items provided when learning is not progressing, and articles A, B, and C are displayed.
  • the UI example 42 shows a display example of items provided when learning progresses, and articles D, E, and F are displayed. For example, when learning progresses and a model having a high score among a plurality of models to be selected is generated, the UI example 42 is generated by the model having a high score.
  • the UI example 41 and the UI example 42 are compared, it can be seen that different items are displayed according to the progress of learning.
  • the fact that “articles that everyone likes to read” is listed is displayed as information indicating that the model is generated using a model that has not been learned.
  • “information recommended for you” is displayed as information indicating that the model has been generated using a model with advanced learning. As described above, in the example illustrated in FIG. 7, different items are displayed according to the progress of learning, and information indicating the progress of learning is displayed.
  • FIG. 8 is a diagram for explaining an example of a UI according to the present embodiment.
  • UI example 51 shows a display example of an item generated using a model for recommended use, and it is displayed on the bat purchase screen that a person who bought the bat is buying another ball or glove.
  • the UI example 52 shows a display example of an item generated using a model for prediction use, and the bat sales prediction is displayed on the bat purchase screen.
  • the UI example 51 is displayed to a user who has a high probability of purchasing on a purchase screen generated using a model for recommended use.
  • the UI example 52 is displayed to a user who has a high probability of purchasing on the purchase screen generated using the model for prediction use.
  • the probability of purchase that is, a model having a high score varies from user to user
  • different items are displayed for each user.
  • the person who bought the bat is buying a ball or a glove as information indicating that the model is generated using a recommended use model having a high score.
  • sales prediction of a bat is displayed as information indicating that a model for prediction use having a high score is used.
  • different items are displayed according to the usage of the model used, and information indicating the usage of the model is displayed.
  • the server 200 may provide information indicating whether or not the feedback tendency changes depending on the context of the user when receiving the information, for example, as the information indicating the generation reason. For example, when the model used to generate the item is contextual BA and the score of contextual BA is high, the server 200 may provide information indicating that the tendency of feedback changes depending on the context. On the other hand, if the model used to generate the item is a non-contextual BA and the score of the non-contextual BA is high, the server 200 provides information indicating that the tendency of feedback does not change depending on the context. Also good. An example of such a UI will be described with reference to FIGS.
  • FIG. 9 is a diagram for explaining an example of a UI according to the present embodiment.
  • UI examples 61 to 63 show conversion candidate display screens when “Yes” is input.
  • the UI example 61 shows a display example of items provided to a user who always creates a sentence in a sentence style regardless of the context.
  • UI examples 62 and 63 are display examples provided to a user who creates a sentence by switching between a sentence form and a colloquial form depending on the context.
  • conversion candidates that give priority to the text style that is, give priority to “Thank you” over “Thank you!” are displayed.
  • the UI example 62 corresponds to a display example of an item generated by a non-contextual model
  • the UI examples 62 and 63 correspond to display examples of an item generated by a contextual model. That is, in the example shown in FIG. 9, different items are displayed depending on whether the model used to generate the item is contextual or not contextual.
  • FIG. 10 is a diagram for explaining an example of a UI according to the present embodiment.
  • the UI example 71 shows a display example of items provided to a user whose reading preference does not always change regardless of the context, and the news that the user likes is in the order of “entertainment”, “sports”, and “international”. Is displayed.
  • the UI example 72 shows a display example of an item provided to a user whose taste of the news to be read changes depending on the context. The news that the user likes is “economy” when on the train, and “ “Sports” and “Lifehack” is displayed before going to bed.
  • the UI example 71 corresponds to a display example of an item generated by a non-contextual model
  • the UI example 72 corresponds to a display example of an item generated by a contextual model. That is, in the example shown in FIG. 10, different items are displayed depending on whether the model used to generate the item is contextual or not contextual. Furthermore, in the UI example 71, “You always like this article” is displayed, indicating that the tendency of feedback does not change depending on the context. Also, in the UI example 72, “You are a person whose preference changes depending on the behavior” indicating that the tendency of feedback changes according to the context is displayed. As described above, in the example illustrated in FIG. 10, information indicating whether or not the feedback tendency changes according to the context of the user when the provision of information is received is displayed.
  • FIG. 11 is a sequence diagram illustrating an example of a flow of information providing processing executed in the information processing system 1 according to the present embodiment. As shown in FIG. 11, the terminal device 100 and the server 200 are involved in this sequence.
  • the terminal device 100 collects information indicating a context (step S102). For example, the terminal device 100 acquires information indicating a user's context by a user input to a touch panel or the like or a sensor device such as a built-in gyro sensor, acceleration sensor, biological information sensor, camera, or microphone.
  • the information indicating the context may be information acquired by a device other than the terminal device 100 such as posting to the SNS, and may be acquired by the server 200, for example.
  • the information indicating the context may include location information of the terminal device 100 specified by GPS, Wi-Fi (registered trademark), Bluetooth (registered trademark), a base station, or the like, or may include time information or the like. You may go out.
  • the terminal device 100 transmits information indicating the context to the server 200 (step S104).
  • the server 200 assigns a plurality of models to be selected to the model selection algorithm (step S106).
  • the model assignment may be performed before receiving information indicating the context.
  • the server 200 selects a model by the model selection algorithm (step S108), and generates an item using the selected model (step S110). And the server 200 transmits the information which shows the produced
  • the terminal device 100 outputs the item received from the server 200 to the user (step S114).
  • the terminal device 100 may display a UI as shown in an example in FIGS. 7 to 10, or may output an item together with the display or instead of the display by voice or vibration.
  • the terminal device 100 acquires feedback from the user for the item, and transmits information indicating the acquired user feedback to the server 200 (step S116).
  • the server 200 performs learning by updating each parameter of the model assigned to the model selection algorithm based on the information indicating the user feedback received from the terminal device 100 (step S118).
  • FIG. 12 is a block diagram illustrating an example of a hardware configuration of the information processing apparatus according to the present embodiment.
  • the information processing apparatus 900 illustrated in FIG. 12 can implement the terminal apparatus 100 or the server 200 illustrated in FIG. 3, for example.
  • Information processing by the terminal device 100 or the server 200 according to the present embodiment is realized by cooperation of software and hardware described below.
  • the information processing device 900 includes a CPU (Central Processing Unit) 901, a ROM (Read Only Memory) 902, a RAM (Random Access Memory) 903, and a host bus 904a.
  • the information processing apparatus 900 includes a bridge 904, an external bus 904b, an interface 905, an input device 906, an output device 907, a storage device 908, a drive 909, a connection port 911, and a communication device 913.
  • the information processing apparatus 900 may include a processing circuit such as a DSP or an ASIC in place of or in addition to the CPU 901.
  • the CPU 901 functions as an arithmetic processing unit and a control unit, and controls the overall operation in the information processing apparatus 900 according to various programs. Further, the CPU 901 may be a microprocessor.
  • the ROM 902 stores programs used by the CPU 901, calculation parameters, and the like.
  • the RAM 903 temporarily stores programs used in the execution of the CPU 901, parameters that change as appropriate during the execution, and the like.
  • the CPU 901 can form the processing unit 230 illustrated in FIG. 3.
  • the CPU 901, ROM 902, and RAM 903 are connected to each other by a host bus 904a including a CPU bus.
  • the host bus 904 a is connected to an external bus 904 b such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge 904.
  • an external bus 904 b such as a PCI (Peripheral Component Interconnect / Interface) bus
  • PCI Peripheral Component Interconnect / Interface
  • the host bus 904a, the bridge 904, and the external bus 904b do not necessarily have to be configured separately, and these functions may be mounted on one bus.
  • the input device 906 is realized by a device in which information is input by the user, such as a mouse, a keyboard, a touch panel, a button, a microphone, a switch, and a lever.
  • the input device 906 may be, for example, a remote control device using infrared rays or other radio waves, or may be an external connection device such as a mobile phone or a PDA that supports the operation of the information processing device 900.
  • the input device 906 may include, for example, an input control circuit that generates an input signal based on information input by the user using the above-described input means and outputs the input signal to the CPU 901.
  • a user of the information processing apparatus 900 can input various data and instruct a processing operation to the information processing apparatus 900 by operating the input device 906.
  • the input device 906 may be formed by a sensor that senses information about the user.
  • the input device 906 includes various sensors such as an image sensor (for example, a camera), a depth sensor (for example, a stereo camera), an acceleration sensor, a gyro sensor, a geomagnetic sensor, an optical sensor, a sound sensor, a distance sensor, and a force sensor. Can be included.
  • the input device 906 includes information related to the information processing device 900 state, such as the posture and movement speed of the information processing device 900, and information related to the surrounding environment of the information processing device 900, such as brightness and noise around the information processing device 900. May be obtained.
  • the input device 906 may include a GPS sensor that receives GPS signals and measures position information such as latitude, longitude, and altitude of the device. As for the position information, the input device 906 may acquire position information measured based on a distance from a base station or access point such as Wi-Fi, Bluetooth, or LTE. The input device 906 may acquire time information from a built-in clock or from another device.
  • the terminal device 100 acquires information indicating the user's context using the input device 906, for example.
  • the output device 907 is formed of a device that can notify the user of the acquired information visually or audibly. Examples of such devices include CRT display devices, liquid crystal display devices, plasma display devices, EL display devices, display devices such as laser projectors, LED projectors and lamps, audio output devices such as speakers and headphones, printer devices, and the like. .
  • the output device 907 outputs results obtained by various processes performed by the information processing device 900. Specifically, the display device visually displays results obtained by various processes performed by the information processing device 900 in various formats such as text, images, tables, and graphs.
  • the audio output device converts an audio signal composed of reproduced audio data, acoustic data, and the like into an analog signal and outputs it aurally.
  • the terminal device 100 outputs an item generated by the server 200 using the display device and the audio output device.
  • the storage device 908 is a data storage device formed as an example of a storage unit of the information processing device 900.
  • the storage apparatus 908 is realized by, for example, a magnetic storage device such as an HDD, a semiconductor storage device, an optical storage device, a magneto-optical storage device, or the like.
  • the storage device 908 may include a storage medium, a recording device that records data on the storage medium, a reading device that reads data from the storage medium, a deletion device that deletes data recorded on the storage medium, and the like.
  • the storage device 908 stores programs executed by the CPU 901, various data, various data acquired from the outside, and the like.
  • the storage device 908 can form the storage unit 220 shown in FIG. 3, for example.
  • the drive 909 is a storage medium reader / writer, and is built in or externally attached to the information processing apparatus 900.
  • the drive 909 reads information recorded on a removable storage medium such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, and outputs the information to the RAM 903.
  • the drive 909 can also write information to a removable storage medium.
  • connection port 911 is an interface connected to an external device, and is a connection port with an external device capable of transmitting data by USB (Universal Serial Bus), for example.
  • USB Universal Serial Bus
  • the communication device 913 is a communication interface formed by a communication device or the like for connecting to the network 920, for example.
  • the communication device 913 is, for example, a communication card for wired or wireless LAN (Local Area Network), LTE (Long Term Evolution), Bluetooth (registered trademark), or WUSB (Wireless USB).
  • the communication device 913 may be a router for optical communication, a router for ADSL (Asymmetric Digital Subscriber Line), a modem for various communication, or the like.
  • the communication device 913 can transmit and receive signals and the like according to a predetermined protocol such as TCP / IP, for example, with the Internet and other communication devices.
  • the communication device 913 can form, for example, the communication unit 210 illustrated in FIG.
  • the network 920 is a wired or wireless transmission path for information transmitted from a device connected to the network 920.
  • the network 920 may include a public line network such as the Internet, a telephone line network, and a satellite communication network, various LANs including the Ethernet (registered trademark), a wide area network (WAN), and the like.
  • the network 920 may include a dedicated line network such as an IP-VPN (Internet Protocol-Virtual Private Network).
  • IP-VPN Internet Protocol-Virtual Private Network
  • each of the above components may be realized using a general-purpose member, or may be realized by hardware specialized for the function of each component. Therefore, it is possible to change the hardware configuration to be used as appropriate according to the technical level at the time of carrying out this embodiment.
  • a computer program for realizing each function of the information processing apparatus 900 according to the present embodiment as described above can be produced and mounted on a PC or the like.
  • a computer-readable recording medium storing such a computer program can be provided.
  • the recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, or the like.
  • the above computer program may be distributed via a network, for example, without using a recording medium.
  • the server 200 provides the user with information generated using one model selected from a plurality of models to be selected by the model selection algorithm, and responds to user feedback on the provided information. Generate information to be provided to the user.
  • the server 200 learns each of a plurality of models prepared in advance based on user feedback, and selects a model with a high score to generate an item, thereby providing the user with more appropriate information. Is possible.
  • a providing unit that provides a user with information generated using a single model selected by a model selection algorithm from a plurality of models to be selected;
  • a generating unit that generates information to be provided to the user in response to user feedback on the information provided by the providing unit;
  • An information processing apparatus comprising: (2) The information processing apparatus according to (1), wherein the providing unit provides information indicating characteristics of a model used for generating information to be provided to a user among the plurality of models to be selected.
  • the information processing apparatus according to any one of (5) to (8), wherein the plurality of models to be selected include models of the same type but different parameters.
  • the information processing apparatus according to any one of (5) to (9), wherein the plurality of models to be selected includes a model selection algorithm that selects a lower model as a selection target.
  • the information processing apparatus according to any one of (5) to (10), wherein the plurality of models to be selected include models with different learning progresses.
  • the information processing apparatus according to any one of (5) to (11), wherein the plurality of models to be selected include models that are commonly included among different users.
  • the information processing apparatus according to any one of (5) to (12), wherein the plurality of models to be selected includes a model that considers a user's context and a model that does not consider the user's context.
  • the information processing apparatus according to any one of (5) to (13), wherein the plurality of models to be selected includes a model corresponding to user attribute information.
  • the assigning unit excludes a model from the selection target or adds a new model according to a score indicating the priority of the plurality of models to be selected in the model selection algorithm, (5) to (14) The information processing apparatus according to any one of the above.
  • the information processing apparatus further includes a learning unit that learns parameters of each of the plurality of models to be selected in response to user feedback with respect to information provided by the providing unit (1) to (15) The information processing apparatus according to any one of the above. (17) The information processing apparatus according to (16), wherein the learning unit performs learning for each user. (18) The information processing apparatus according to (17), wherein the learning unit reflects a learning result based on feedback of one user also for other users with respect to a model included in common among different users.
  • An information processing method including: (20) Computer A providing unit that provides a user with information generated using a single model selected by a model selection algorithm from a plurality of models to be selected; A generating unit that generates information to be provided to the user in response to user feedback on the information provided by the providing unit; Program to function as.
  • Model 20 Model selection algorithm 100 Terminal device 200 Server 210 Communication part 220 Storage part 230 Processing part 231 Allocation part 232 Generation part 233 Provision part 234 Learning part

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention concerne un dispositif de traitement d'informations comprenant : une unité de fourniture qui fournit, à un utilisateur, des informations générées au moyen d'un modèle unique sélectionné parmi une pluralité de modèles candidats de sélection par un algorithme de sélection de modèle ; et une unité de génération qui génère des informations qui doivent être fournies à l'utilisateur en fonction d'une rétroaction fournie par l'utilisateur pour les informations fournies par l'unité de fourniture.
PCT/JP2016/054961 2015-05-27 2016-02-19 Dispositif de traitement d'informations, procédé de traitement d'informations et programme WO2016189905A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2015107819 2015-05-27
JP2015-107819 2015-05-27

Publications (1)

Publication Number Publication Date
WO2016189905A1 true WO2016189905A1 (fr) 2016-12-01

Family

ID=57393111

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2016/054961 WO2016189905A1 (fr) 2015-05-27 2016-02-19 Dispositif de traitement d'informations, procédé de traitement d'informations et programme

Country Status (1)

Country Link
WO (1) WO2016189905A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2018142295A (ja) * 2017-02-28 2018-09-13 富士ゼロックス株式会社 機械学習に基づくアクセス制御のためのシステムおよび方法、プログラム
WO2019171677A1 (fr) * 2018-03-08 2019-09-12 オムロン株式会社 Dispositif de sortie de données, procédé de sortie de données, et programme
CN112632179A (zh) * 2019-09-24 2021-04-09 北京国双科技有限公司 模型构建方法、装置、存储介质及设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005091214A1 (fr) * 2004-03-18 2005-09-29 Denso It Laboratory, Inc. Systeme de traitement d’informations sur un vehicule, procede de traitement d’informations sur un vehicule et programme
JP2007058398A (ja) * 2005-08-23 2007-03-08 Kddi Corp コンテンツ推薦装置、コンテンツ推薦方法及びコンピュータプログラム
JP2009140075A (ja) * 2007-12-04 2009-06-25 Sony Corp 情報処理装置および方法、並びにプログラム
JP2010044518A (ja) * 2008-08-11 2010-02-25 Kddi Corp 嗜好推定装置、嗜好推定方法及びコンピュータプログラム
JP2011511968A (ja) * 2007-10-04 2011-04-14 キシアム・テクノロジーズ・リミテッド リコメンデーション生成システム、装置、及び方法
JP2012118572A (ja) * 2010-11-29 2012-06-21 Nec Corp コンテンツ推薦システム、コンテンツ推薦装置、推薦方式制御方法、及び推薦方式制御プログラム

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005091214A1 (fr) * 2004-03-18 2005-09-29 Denso It Laboratory, Inc. Systeme de traitement d’informations sur un vehicule, procede de traitement d’informations sur un vehicule et programme
JP2007058398A (ja) * 2005-08-23 2007-03-08 Kddi Corp コンテンツ推薦装置、コンテンツ推薦方法及びコンピュータプログラム
JP2011511968A (ja) * 2007-10-04 2011-04-14 キシアム・テクノロジーズ・リミテッド リコメンデーション生成システム、装置、及び方法
JP2009140075A (ja) * 2007-12-04 2009-06-25 Sony Corp 情報処理装置および方法、並びにプログラム
JP2010044518A (ja) * 2008-08-11 2010-02-25 Kddi Corp 嗜好推定装置、嗜好推定方法及びコンピュータプログラム
JP2012118572A (ja) * 2010-11-29 2012-06-21 Nec Corp コンテンツ推薦システム、コンテンツ推薦装置、推薦方式制御方法、及び推薦方式制御プログラム

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2018142295A (ja) * 2017-02-28 2018-09-13 富士ゼロックス株式会社 機械学習に基づくアクセス制御のためのシステムおよび方法、プログラム
JP7056049B2 (ja) 2017-02-28 2022-04-19 富士フイルムビジネスイノベーション株式会社 機械学習に基づくアクセス制御のためのシステムおよび方法、プログラム
WO2019171677A1 (fr) * 2018-03-08 2019-09-12 オムロン株式会社 Dispositif de sortie de données, procédé de sortie de données, et programme
JP2019159435A (ja) * 2018-03-08 2019-09-19 オムロン株式会社 データ出力装置、データ出力方法及びプログラム
CN112632179A (zh) * 2019-09-24 2021-04-09 北京国双科技有限公司 模型构建方法、装置、存储介质及设备

Similar Documents

Publication Publication Date Title
JP6558364B2 (ja) 情報処理装置、情報処理方法及びプログラム
US10733255B1 (en) Systems and methods for content navigation with automated curation
US9026941B1 (en) Suggesting activities
US20220237486A1 (en) Suggesting activities
KR102331049B1 (ko) 통신 개시를 위한 사용자 신호 레버리징
US9058563B1 (en) Suggesting activities
US20170154626A1 (en) Question and answer processing method and electronic device for supporting the same
US10911807B2 (en) Optimization of an automation setting through selective feedback
US20190147841A1 (en) Methods and systems for displaying a karaoke interface
US11720640B2 (en) Searching social media content
US11436265B2 (en) System for presenting tailored content based on user sensibilities
JP2018129004A (ja) 生成装置、生成方法、及び生成プログラム
JP6543774B2 (ja) コンピューティングデバイスのプロパティに基づく情報のランク付け
US20200110788A1 (en) Mediating method and device
KR20180060971A (ko) 사용자 입력에 기반한 문장을 제공하는 장치 및 방법
KR20140113436A (ko) 관계 모델 매커니즘을 구현한 컴퓨팅 시스템 및 그 작동 방법
US20190147229A1 (en) Methods and systems for playing musical elements based on a tracked face or facial feature
CN113939870A (zh) 为基于生成和检索的应用学习有效音乐特征的技术
WO2016189905A1 (fr) Dispositif de traitement d'informations, procédé de traitement d'informations et programme
JP6733662B2 (ja) 情報処理装置、情報処理方法及びコンピュータプログラム
JPWO2018092226A1 (ja) コンピュータシステム、ａｐｉ提供方法及びプログラム
JP7315705B2 (ja) タッチパッド入力を備えたゲームコントローラ
JP2015152940A (ja) 提示制御装置、提示制御方法、およびプログラム
WO2019073668A1 (fr) Dispositif de traitement d'informations, procédé de traitement d'informations et programme
WO2023015079A1 (fr) Espace de travail orienté humeur

Legal Events

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

Ref document number: 16799614

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16799614

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: JP