Information processing device, information processing method, and information processing program

Abstract

[Problem] A search using a trained AI model infers an answer to a question which is a prompt, so that a result of the inference is output as a search result. Thus, there is a problem where, unlike a search using a search engine where a search result is obtained by entering a keyword in a search box for each item, it is impossible to revise individual search conditions or to guarantee strict search conditions. [Solution] To solve the problem, the present invention provides an information processing device and the like having a search instruction setting unit for setting a search instruction in a natural language, and a search condition generation unit for inputting the set search instruction to a trained AI model and generating individual corresponding search conditions.

Description

Information Processing Apparatus, Information Processing Method, and Information Processing Program

[0001] The present invention relates to an information processing apparatus that searches for information.

[0002] In recent years, the use of large language models (hereinafter sometimes abbreviated as LLMs) has been spreading. LLMs can execute various natural language processing tasks such as text classification, information extraction, text summarization, text generation, and question answering. For example, Patent Document 1 discloses a technique for extracting predetermined information from various printed materials such as advertising flyers using an LLM.

[0003] Japanese Patent No. 7430437

[0004] By giving an LLM a prompt that indicates the target to be extracted, as in the technique of Patent Document 1, predetermined information is answered as an extraction result. The prompt is in natural language. For example, by giving an LLM a sentence such as "Please show the 1:10,000 topographic map of Setagaya Ward.", a topographic map that meets the conditions is returned.

[0005] As described above, the search using an LLM infers an answer to a question called a prompt and outputs the inference result as a search result. Therefore, there is a problem that it is impossible to modify search conditions or ensure strict search conditions as in a search by a search engine that obtains search results by inputting keywords in a search box for each item.

[0006] Therefore, in order to solve the above problems, the present invention provides the following information processing apparatus and the like. That is, an information processing apparatus having a search instruction setting unit that sets a search instruction in natural language, and a search condition generation unit that inputs the set search instruction into a learned AI model to generate corresponding individual search conditions is provided.

[0007] Further, in addition to the above features, an information processing apparatus further having a search condition setting unit that sets the generated search conditions on a UI screen is provided.

[0008] Furthermore, in addition to the above features, the present invention provides an information processing device that further includes a corresponding location display unit that displays the corresponding location in the set search instruction for each of the set search conditions.

[0009] Furthermore, in addition to the above features, the present invention provides an information processing device that further includes a cursor movement unit that moves the cursor to an area on the UI screen where search conditions corresponding to the displayed area are set, based on an instruction operation on the displayed area.

[0010] Furthermore, the present invention provides an information processing method executed by an information processing device, comprising: a search instruction setting step of setting search instructions in natural language; and a search condition generation step of inputting the set search instructions into a trained AI model to generate corresponding individual search conditions.

[0011] Furthermore, the present invention provides an information processing program that causes an information processing device to execute a search instruction setting step, which involves setting search instructions in natural language, and a search condition setting step, which involves inputting the set search instructions into a trained AI model to generate corresponding individual search conditions.

[0012] The present invention provides an information processing device that enables search instructions in natural language while also allowing for modification of search conditions and ensuring strict adherence to those conditions.

[0013] A block diagram showing an example of the functional configuration of the information processing device of the embodiment. A conceptual diagram showing an example of displaying search instructions and search conditions on the UI screen. A conceptual diagram explaining the relevant location display unit and cursor movement unit. A conceptual diagram showing an example of the hardware configuration for realizing the information processing device of the embodiment. A flowchart showing a simplified example of the processing flow of the information processing device of the embodiment.

[0014] Embodiments of the present invention will be described below with reference to the accompanying drawings. However, the present invention is not limited in any way to these embodiments, and can be implemented in various ways without departing from its essence.

[0015] <Examples> <Summary> The present invention is characterized by generating corresponding individual search conditions from search instructions set in natural language when searching for information from various sources. This allows for modification of the generated search conditions.

[0016] The following describes the functions and processing flow of the information processing device, as well as the hardware components. The functional blocks of this system described below can be implemented as a combination of hardware and software. Specifically, if a computer is used, these may include hardware components such as a CPU (Central Processing Unit), main memory, a bus, or secondary storage devices (hard disk drives, non-volatile memory, storage media such as CDs and DVDs, and their readers), input devices used for information input, printing equipment, display devices, and other external peripheral devices, as well as interfaces for these external peripheral devices, communication interfaces, driver programs and other application programs for controlling the hardware, and user interface applications. The CPU's arithmetic processing, based on programs deployed in main memory, processes and stores data input from input devices and other interfaces, and stored in memory and on the hard disk, and generates instructions for controlling the aforementioned hardware and software. Alternatively, the functional blocks of this system may be implemented using dedicated hardware.

[0017] Furthermore, this invention can be realized not only as a system but also as a method. Moreover, a part of such an invention can be configured as software. In addition, programs used to cause a computer to execute such software, and recording media on which such programs are fixed, are naturally included within the technical scope of this invention (as is the case throughout this specification).

[0018] <Functional Configuration> Figure 1 is a block diagram showing an example of the functional configuration of the information processing device of this embodiment. As shown in Figure 1, the information processing device 100 includes a search instruction setting unit 101, a search condition generation unit 102, a search condition setting unit 103, a corresponding location display unit 104, and a cursor movement unit 105.

[0019] In this embodiment, the trained AI model is a machine learning model that has been pre-trained and is capable of handling general-purpose tasks. A typical example is a large-scale language model, but it may also be a small-scale language model or a multimodal language model that can handle images or other formats.

[0020] <Search Instruction Setting Unit> The search instruction setting unit 101 has the function of setting search instructions in natural language. Search instructions in natural language are linguistic expressions that specify the object to be searched for or request that the object be presented, and are especially intended to be spoken language that people normally use. This is intended to allow prompts (interactive instructions or questions) to be input to the trained AI model when using the trained AI model for searching.

[0021] The search instruction may be set, for example, by the user entering a sentence instructing the search target into an input box, or by voice input by the user. Alternatively, the information processing device may be configured to present a predetermined search instruction to the user, and the user may modify the presented search instruction before setting it as the search instruction.

[0022] <Search Condition Generation Unit> The search condition generation unit 102 has the function of inputting the set search instructions into the trained AI model and generating corresponding individual search conditions. Search conditions are the conditions specified when performing a search. For example, when searching for restaurants, the search conditions would be conditions that specify search items such as "location or station name," "genre," and "scene."

[0023] For example, if the set search instruction is "Please recommend a restaurant near Shibuya Station that serves Italian food and is suitable for a banquet," inputting this search instruction into a trained AI model will allow it to infer the individual search conditions included in the instruction and generate search conditions such as "Location or station name = Shibuya Station," "Genre = Italian cuisine," and "Scene = Banquet." Similarly, inputting the search instruction "Please provide a 1:10,000 topographic map of Setagaya Ward" into the trained AI model will generate search conditions such as "Region = Setagaya Ward," "Scale = 1:10,000," and "Type = Topographic map."

[0024] Furthermore, search conditions can be generated not only by directly generating them from the set search instructions, but also by selecting appropriate search conditions from those pre-configured in the source being searched (database, server, etc.). For example, in a bridge database, if search items such as "bridge type classification," "bridge type," "road surface type," "deck type," and "construction method" are pre-configured, the corresponding search items can be selected from the set search instructions, and those search conditions can be generated.

[0025] <Search Condition Setting Unit> The search condition setting unit 103 has the function of setting the generated search conditions on the UI screen. Figure 2 is a conceptual diagram showing an example of displaying search instructions and search conditions on the UI screen. Figure 2(a) shows the case when a search instruction is entered. As shown in the figure, the UI (User Interface) screen 201 has a field 202 for entering search instructions, and the text "What map are you looking for?" is displayed as guidance.

[0026] In this field, the search instruction "Please provide a 1:10,000 topographic map of Setagaya Ward" has been entered. On the left side of the UI screen, a search criteria field 203 is displayed for entering search conditions. In this example, the search criteria fields for each of the pre-set search items, "Region," "Type," "Scale," and "Display Items," are displayed. Further down, a clear button 204 for clearing the search conditions and a search button 205 for performing a search based on the set search conditions are displayed.

[0027] Once the search instruction is confirmed, as shown in Figure 2(b), the input for each search condition field is automatically performed, and the search conditions "Setagaya Ward" is displayed for "Region," "Topographic Map" for "Type," and "1 / 10,000" for "Scale," while "Do not specify" is displayed for "Display items."203 This is because the trained AI model generates search conditions corresponding to the pre-set search items from the entered search instruction, "Please display a 1:10,000 topographic map of Setagaya Ward," and the search condition setting unit displays these search conditions on the UI screen. The user can also delete or modify the displayed search conditions.

[0028] By entering search instructions in this way, each search condition is displayed on the UI screen, making it clear how the search instructions have been reflected in the search conditions, and allowing for modifications to the search conditions. Furthermore, by setting the original search instructions again, the search conditions can be regenerated.

[0029] <Relevant Location Display Unit> The relevant location display unit 104 has the function of displaying the relevant location in the input search instruction for each of the set search conditions. This clarifies the correspondence between each search condition and the relevant location in the search instruction.

[0030] Figure 3 is a conceptual diagram illustrating the display area and cursor movement area. As shown in Figure 3(a), the input field 302 for the search instruction on the UI screen 301 displays the search instruction, "Please display the 1:10,000 topographic map of Setagaya Ward." The words "Setagaya Ward," "1:10,000," and "topographic map" are enclosed in square brackets and underlined. These words are search conditions for "region," "type," and "scale," respectively. By displaying it in this way, it is possible to indicate which words of the natural language search instruction were used as search conditions.

[0031] Furthermore, the system can be configured to clearly indicate which words in the search instruction are used in which search conditions. As shown in Figure 3(b), by pointing the cursor 304 at the words "Setagaya Ward" in the search instruction input field 302 on the UI screen 301, "Please provide a 1:10,000 topographic map of Setagaya Ward," both the words "Setagaya Ward" and the words "Setagaya Ward" 305 in the corresponding search condition field "Region" are highlighted by inverting the colors. This makes the correspondence between individual search conditions and the corresponding locations in the search instructions clear. Note that various highlighting methods are possible, not limited to inversion of colors.

[0032] Conversely, the system can also be configured so that the relevant section of the search instruction is highlighted by pointing a cursor or the like at any of the search criteria in the search criteria field 303. Furthermore, voice input may also be used.

[0033] <Cursor Movement Unit> The cursor movement unit 105 has the function of moving the cursor to an area on the UI screen where the search conditions corresponding to the displayed location are set, based on an instruction operation for that location.

[0034] For example, as shown in Figure 3(b), when the cursor 304 is pointing to the word "Setagaya-ku" in the search instruction, performing an operation such as a left click, which is a selection or execution action, moves the cursor 304 to the "Region" field of the search condition field 303. The cursor may be moved while maintaining the arrow-shaped mouse cursor form as shown in the figure, or it may be moved to a text cursor form and positioned at the end of the word "Setagaya" in the "Region" field, making it immediately possible to input text.

[0035] By configuring the system to allow this type of cursor movement, it becomes easy and quick to switch between the search criteria and the words used in the search instructions, enabling rapid and reliable review and modification of the search criteria.

[0036] <Hardware Configuration> Figure 4 is a conceptual diagram showing an example of the hardware configuration for realizing the information processing device according to the embodiment. As shown in the figure, the information processing device 400 includes a CPU 401 that performs various calculations, a RAM 402 which is a volatile recording medium, a storage device 403 such as a flash memory or HDD which is a non-volatile storage medium, a communication interface 404, and an input / output interface 405. The RAM 402 reads programs that perform various calculations for the CPU 401 to execute and provides a work area (work area) for those programs. In addition, multiple addresses are assigned to the RAM 402, and programs executed by the CPU 401 can exchange data with each other and perform processing by identifying and accessing these addresses (this is the same throughout this specification).

[0037] Here, the functions of the search instruction setting unit 101, search condition generation unit 102, and search condition setting unit 103 of the information processing device 100 in Figure 1 are mainly realized by the CPU 401 and RAM 402 in Figure 4. In addition, the functions of the relevant location display unit 104 and cursor movement unit 105 are mainly realized by the CPU 401, RAM 402, and input / output interface 405 in Figure 4. Furthermore, if the search condition generation unit 102 uses an externally existing trained AI model, each function is realized by exchanging signals and information with each other via the communication interface 404 and input / output interface 405.

[0038] <Processing Flow> Figure 5 is a simplified flowchart showing an example of the processing flow of the information processing device of Embodiment 1. First, a search instruction is set in natural language (S501: Search Instruction Setting Step). Then, the set search instruction is input to the trained AI model to generate corresponding individual search conditions (S502: Search Condition Generation Step). Then, the generated search conditions are set on the UI screen (S503: Search Condition Setting Step). Then, for each of the set search conditions, the corresponding part in the set search instruction is displayed (S504: Corresponding Part Display Step). Then, by performing an instruction operation on the displayed corresponding part, the cursor is moved to the area on the UI screen where the corresponding search condition is set (S505: Cursor Movement Step).

[0039] <Effects> According to the information processing device of this embodiment, it is possible to modify individual search conditions generated from search instructions in natural language.

[0040] 100, 400: Information processing device 101: Search instruction setting unit 102: Search condition generation unit 103: Search condition setting unit 104: Corresponding location display unit 105: Cursor movement unit 401: CPU 402: RAM 403: Storage 404: Communication interface 405: Input / output interface

Claims

1. An information processing device comprising: a search instruction setting unit that sets search instructions in natural language; and a search condition generation unit that inputs the set search instructions into a trained AI model and generates corresponding individual search conditions.

2. The information processing apparatus according to claim 1, further comprising a search condition setting unit for setting the generated search conditions on a UI screen.

3. The information processing apparatus according to claim 2, further comprising a corresponding location display unit that displays the corresponding location in the set search instruction for each of the set search conditions.

4. The information processing apparatus according to claim 3, further comprising a cursor movement unit that moves the cursor to an area on the UI screen where search conditions corresponding to the displayed area are set, based on an instruction operation on the displayed area.

5. An information processing method to be executed by an information processing device, comprising: a search instruction setting step of setting a search instruction in natural language; and a search condition generation step of inputting the set search instruction into a trained AI model to generate corresponding individual search conditions.

6. An information processing program that causes an information processing device to execute a search instruction setting step, which sets a search instruction in natural language, and a search condition setting step, which inputs the set search instruction into a trained AI model to generate corresponding individual search conditions.

Images