Program, information processing device, and information processing method

Abstract

To display more relevant search results based on the user's level. [Solution] The program of the embodiment causes a computer equipped with a storage device that stores content and the level of the content to function as a query reception unit, a user level determination unit, a search unit, and a display control unit. The query reception unit receives a first query used to search for the content. The user level determination unit determines a user level that indicates the level of at least one skill of the user. The search unit searches for the content based on the first query and the user level. The display control unit displays display information including at least one of the searched contents on a display device.

Description

Technical Field

[0001] Embodiments of the present invention relate to a program, an information processing apparatus, and an information processing method.

Background Art

[0002] Conventionally, there has been known a technique for estimating the level of content and displaying the content in accordance with search results or recommending content at a level equivalent to the level of search results in order to assist a user in selecting content.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the conventional technology, it has been difficult to display more appropriate search results of content according to the user level.

Means for Solving the Problems

[0005] The program according to the embodiment causes a computer including a storage device storing content and the level of the content to function as a query reception unit, a user level determination unit, a search unit, and a display control unit. The query reception unit receives a first query used for searching the content. The user level determination unit determines a user level indicating the level of at least one skill of the user. The search unit searches the content based on the first query and the user level. The display control unit displays display information including at least one of the searched content on a display device.

Brief Description of the Drawings

[0006] [Figure 1] A diagram showing an example of the functional configuration of the information processing device according to the first embodiment. [Figure 2] A diagram showing an example of a keyword list for the first embodiment. [Figure 3] A diagram showing an example of display information in the first embodiment. [Figure 4] A flowchart illustrating an example of the information processing method of the first embodiment. [Figure 5] A diagram showing an example of the functional configuration of the information processing device according to the second embodiment. [Figure 6] A flowchart illustrating an example of the information processing method of the second embodiment. [Figure 7] A diagram showing an example of the functional configuration of the information processing device according to the third embodiment. [Figure 8] A diagram showing an example of the functional configuration of the information processing device according to the fourth embodiment. [Figure 9] A diagram illustrating example 1 of the adjustment of the solution based on the course history in the fourth embodiment. [Figure 10] A diagram illustrating example 2 of adjusting the solution based on the course history in the fourth embodiment. [Figure 11] A diagram illustrating example 3 of the adjustment of the solution based on the course history in the fourth embodiment. [Figure 12] A diagram illustrating example 4 of the adjustment of the solution based on the course history in the fourth embodiment. [Figure 13] A flowchart illustrating an example of the information processing method of the fourth embodiment. [Figure 14A] A diagram showing an example of the functional configuration of the information processing device according to the fifth embodiment. [Figure 14B] A diagram showing an example of the functional configuration of the information processing device according to the fifth embodiment. [Figure 15] A diagram illustrating the user-level-based weight calculation process of the fifth embodiment. [Figure 16A] A flowchart illustrating an example of the information processing method according to the fifth embodiment. [Figure 16B] A flowchart illustrating an example of the information processing method according to the fifth embodiment. [Figure 17]A diagram showing an example of the functional configuration of the information processing apparatus according to the sixth embodiment. [Figure 18] A diagram for explaining Example 1 of the content adjustment process based on the learning history according to the sixth embodiment. [Figure 19] A diagram for explaining Example 2 of the content adjustment process based on the learning history according to the sixth embodiment. [Figure 20] A diagram showing Example 1 of the display information according to the sixth embodiment. [Figure 21] A diagram showing Example 2 of the display information according to the sixth embodiment. [Figure 22] A flowchart showing an example of the information processing method according to the sixth embodiment. [Figure 23] A diagram showing an example of the functional configuration of the information processing apparatus according to the seventh embodiment. [Figure 24] A diagram showing an example of the functional configuration of the information processing apparatus according to the eighth embodiment. [Figure 25] A diagram showing Example 1 of the display information according to the eighth embodiment. [Figure 26] A diagram showing Example 2 of the display information according to the eighth embodiment. [Figure 27A] A diagram showing an example of the functional configuration of the information processing apparatus according to the ninth embodiment. [Figure 27B] A diagram showing an example of the functional configuration of the information processing apparatus according to the ninth embodiment. [Figure 27C] A diagram showing an example of the functional configuration of the information processing apparatus according to the ninth embodiment. [Figure 28] A diagram showing an example of the apparatus configuration of the information processing apparatus according to the first to ninth embodiments.

Embodiments for Carrying Out the Invention

[0007] Hereinafter, embodiments of a program, an information processing apparatus, and an information processing method will be described in detail with reference to the accompanying drawings.

[0008] In the following embodiments, lesson materials provided by educational services such as schools, lifelong learning programs, or workplace training are used as content. In recent years, online classes have been implemented in addition to face-to-face classes, and the content of these classes is recorded and can be viewed on demand. Furthermore, in e-learning, lesson materials are sometimes displayed on the screen, and tests related to the lesson content are conducted along the way. Given this background, in the following embodiments, the materials are not limited to documents such as textbooks, but also include audio and images from educational videos. However, the embodiments are not limited to the examples shown below.

[0009] (First Embodiment) Figure 1 is a diagram showing an example of the functional configuration of the information processing device 1 of the first embodiment. The information processing device 1 of the first embodiment includes a storage control unit 100, a query reception unit 101, a user level determination unit 102, a coupling unit 103, a search unit 104, a display control unit 105, and a storage unit 106.

[0010] The memory control unit 100 controls the storage of data stored in the memory unit 106.

[0011] The memory unit 106 stores a database in which content is stored. Each piece of content is assigned a level. For example, the levels may be pre-set by the memory control unit 100.

[0012] Furthermore, for example, the memory control unit 100 can automatically set the level using a predetermined estimation method or the like based on the text information assigned to the content.

[0013] For content that is originally text-based, such as textbooks, only the text information is retrieved. For audio, text information is obtained through speech recognition and transcription. For images, text information is obtained using OCR (Optical Character Recognition), etc. In addition, content may have titles or descriptions, and these titles or descriptions are also retrieved as text information.

[0014] This document describes how to automatically determine the content level from text information. Several methods can be considered for determining the level. One method is to prepare a keyword list and determine the level based on that list.

[0015] Figure 2 shows an example of a keyword list in the first embodiment. For example, keywords corresponding to each level are defined as shown in Figure 2. The memory control unit 100 assigns the corresponding level if the text information of the content contains the keywords in Figure 2. For example, if the title of the content is "Basic Manager Course", the level "Beginner (1)" is assigned. Also, for example, if the title of the content is "Critical Thinking Application", the level "Advanced (3)" is assigned.

[0016] Another method, for example, is to determine the level using the length of the content. Assume that thicker textbooks are of a higher level, thinner textbooks are of a lower level, and that texts with more characters are of a higher level.

[0017] Another method for determining the level is to use the generality of words in the text information. The more rare and uncommon words a content uses, the higher its level is considered to be, while content with many common words is considered lower in level.

[0018] Specifically, in order to determine the level of content to be stored in the storage unit 106, the memory control unit 100 counts the number of documents containing each word from a large number of pre-prepared documents and calculates the IDF using the following formula. IDF = (Total number of documents) / (Number of documents containing the word)

[0019] The smaller the IDF, the more general the word is included in various articles. The memory control unit 100 calculates the IDF for each word in the text information of each content. Then, the memory control unit 100 uses the maximum value or average value of the IDF as the IDF of the content. The memory control unit 100 increases the level for the content with a higher IDF.

[0020] Also, there is a method using the co-occurrence of words in the text information. The co-occurrence is the ratio used in the same document. Words related to similar topics have a high co-occurrence. For example, (calculus, linear algebra) has a high co-occurrence, but (calculus, curry rice) has a low co-occurrence. The wider the topic range of the content, the lower the co-occurrence between each word, and the narrower the topic range of the content, the higher the co-occurrence. Also, the more specialized the word, the more it is only used in a specific topic, so the co-occurrence between words is higher.

[0021] The memory control unit 100 calculates the co-occurrence rate, for example, by the following formula. Co-occurrence rate = (Number of documents containing word A, word B) / max(Number of documents containing word A, Number of documents containing word B)

[0022] max(A, B) in the above formula is a function that outputs the larger value of A or B.

[0023] For example, the co-occurrence rate of "story" and "tapioca" is obtained as follows. Number of documents containing "story", "tapioca": 50 Number of documents containing "story": 1000 Number of documents containing "tapioca": 80 Co-occurrence rate = 50 / 1000 = 0.05

[0024] As mentioned above, the word "talk" is frequently used and therefore co-occurs with various words, but its co-occurrence with each word is not high. To accurately evaluate the co-occurrence rate with such frequently used words, the co-occurrence rate is calculated by dividing by the more frequent word. The co-occurrence rate can be determined for each word pair in the text information of each content. The maximum and average values ​​of the co-occurrence rates are then used as the co-occurrence level of the document. The memory control unit 100 assigns a higher level to content with a higher co-occurrence rate.

[0025] The memory control unit 100 determines the level numerically using the methods described above. Note that the level may be determined using any one of the methods, or the average of the levels output by each method may be taken.

[0026] The memory unit 106 stores both the content information and the content level together.

[0027] The query reception unit 101 receives text related to the user's request. For example, the text could be a name like "critical thinking" or "accounting," or it could be a sentence like "I want to improve my presentation skills."

[0028] The user level determination unit 102 determines a user level that indicates the level of at least one skill of the user. The user level is determined from various pieces of information. For example, course history can be used as the user level. If there is any content in the storage unit 106 that the user has completed, the user level determination unit 102 assumes that the user has reached the level of that content.

[0029] For example, the user level determination unit 102 calculates the level of proficiency based on the results of the comprehension test after the course.

[0030] Alternatively, the user level determination unit 102 may create comprehension tests from each piece of content even without a learning history, and measure the user level based on the scores of these comprehension tests. In the case of a school, tests are administered to students regularly, so the user level can be measured based on the test results. Another method is to use the user's profile information. If the user is enrolled in a school, information such as grade level and major can be used; if the user is employed by a company, information such as years of employment, position, and job type can be used.

[0031] Furthermore, the user level determination unit 102 generates text related to the user level based on the user level. The following is an example of text generated by the user level determination unit 102. "I have completed courses such as 'Basic Critical Thinking,' 'Basic Accounting,' 'Intermediate Accounting,' and 'Business Etiquette.'" "I have completed the following courses: 'Basic Critical Thinking' (95 points), 'Basic Accounting' (100 points), 'Intermediate Accounting' (80 points), and 'Business Etiquette' (99 points)..." "I'm in my 40s, a section manager, in sales, and I've been with the company for 23 years." "I have a Grade Pre-2 Eiken (English Proficiency Test) and a Grade 2 Information Processing certification."

[0032] Furthermore, the contents of the memory unit 106 may be pre-grouped. For example, the grouping may be determined manually.

[0033] Alternatively, for example, the memory control unit 100 may pre-convert the content into feature vectors and group them using an unsupervised clustering method such as k-means. k-means does not require cluster labels, and when the number of clusters is specified, the content is divided into that number of clusters. A pre-trained clustering model may be used for the clustering model, or if there is a large amount of content, the clustering model may be trained using the features of the content.

[0034] The user level determination unit 102 may calculate a level for each group using the grouped content, the content level, and the learning history. For example, the user level determination unit 102 may determine the user level based on the percentage of content within each group that has been completed. Alternatively, the user level determination unit 102 may sort the content within each group by level and determine a higher user level for users who have completed higher-level content.

[0035] By using this grouping method, it is possible to generate a concise text like the one below without having to describe the entire course history. "My communication skills are 6 out of 10, my management skills are 6 out of 10, and my accounting skills are 10 out of 10."

[0036] The merging unit 103 combines the query entered by the user (first query) with the user level determined by the user level determination unit 102. The merging method is, for example, to combine the two sentences as they are, as shown in Figure 1. That is, the merging unit 103 generates the second query by combining the sentence indicating the user level with the sentence indicating the first query.

[0037] The search unit 104 uses the text generated by the merging unit 103 as a search query (second query) to search the content stored in the memory unit 106. Several search methods are possible.

[0038] One method is keyword search, which searches for content in the memory unit 106 using words in the search query. Stop words are defined in advance. Stop words are general words that are not needed for the search. The search query is split into a sequence of words, and the stop words are excluded. The remaining words are used as keywords, and content containing the keywords is output as search results. If there are multiple keywords, content containing all of them or content containing any one of the keywords can be used, and the search results can also be scored based on the number and frequency of the keywords contained.

[0039] Furthermore, instead of treating each keyword equally, it is possible to prioritize rarer keywords. For example, BM25 is a search method based on the rarity of words. Each keyword in the search query q i In document d of the content, calculate the following formula (1).

[0040]

number

[0041] Here, f(q i ,d) is q in document d i The frequency of each keyword q in the search query is given by: mean(dl) is the average number of words in all documents, |d| is the number of words in document d, and k and b are constants. i The sum of the values ​​calculated for each factor is called the BM25 score. The search unit 104 outputs content with a high BM25 score as search results.

[0042] The search unit 104 may also search for content using vector similarity. Specifically, the search unit 104 first converts the text information obtained from each content into a multi-dimensional vector. Several methods for converting to hectors are possible.

[0043] For example, there is a method called Bag of Words. In Bag of Words, a large number of documents are prepared in advance, and a word list is created from these documents. For the text information of each content, a vector with the dimension of the number of words in the word list is prepared, and the number of words in the text information is used as the value of each element of the vector. Alternatively, the value of each element of the vector can be the number of words multiplied by the IDF of each word.

[0044] Alternatively, a pre-trained machine learning model such as a Transformer may be used for vector transformation. In this case, the search unit 104 transforms the search query into a vector, similar to the content, and calculates the similarity between the search query vector and the vector of each content. For example, the similarity is calculated based on cosine similarity and the distance between the vectors. For example, the search unit 104 outputs the search results in order from the content with the highest similarity. Alternatively, the search unit 104 may set a threshold for similarity and output only content above that threshold, or output only the top few content items.

[0045] The display control unit 105 displays the display information, including the search results, on a display device such as a display.

[0046] Figure 3 shows an example of display information according to the first embodiment. In the example in Figure 3, the left side displays a text box entered by the user and a response statement from the system (information processing device 1). The right side displays the content of the search results. For example, the title and summary of the content are displayed. Since each of the above search methods can generate a search score, the search score is also displayed.

[0047] Figure 4 is a flowchart showing an example of an information processing method according to the first embodiment. The explanation will be based on the flowchart in Figure 4. When a user uses the system (information processing device 1, hereinafter the same), they log in, and the user level determination unit 102 is set up to determine the user's user level.

[0048] First, the search unit 104 generates a search query from the text generated by the aforementioned merging unit 103 (step S101). For example, the merging unit 103 generates a text by combining the text entered by the user in the text box with the text related to the user level generated by the aforementioned user level determination unit 102.

[0049] Next, the search unit 104 performs a search based on the text generated in step S101 (step S102). Finally, the display control unit 105 displays the search results on the display device (step S103).

[0050] As described above, in the information processing device 1 of the first embodiment, the query reception unit 101 receives a first query used to search for content. The user level determination unit 102 determines a user level indicating the level of at least one skill of the user. The search unit 104 searches for content based on the first query and the user level. Then, the display control unit 105 displays display information including at least one searched content on the display device.

[0051] As a result, according to the information processing device 1 of the first embodiment, search results for more appropriate content can be displayed according to the user level.

[0052] (Second Embodiment) Next, a second embodiment will be described. In the description of the second embodiment, explanations similar to those of the first embodiment will be omitted, and the differences from the first embodiment will be described.

[0053] Using user-entered queries directly in a search may not always yield effective results. For example, if a user enters a vague query such as "I want to become a competent worker," it can be difficult to find content similar to that query. Therefore, in the second embodiment, we will describe an embodiment that generates a solution for the input query and searches for content similar to that solution. For example, in the above example, a solution such as "To become a competent worker, you need to be able to plan your work, think logically, communicate, and manage stress" is generated, and content related to planning your work, thinking logically, communication skills, and stress management can be searched.

[0054] Figure 5 shows an example of the functional configuration of the information processing device 1-2 of the second embodiment. The information processing device 1-2 of the second embodiment includes a storage control unit 100, a query receiving unit 101, a user level determination unit 102, a coupling unit 103, a search unit 104, a display control unit 105, a storage unit 106, and a solution generation unit 107. In the second embodiment, a solution generation unit 107 is further added.

[0055] The solution generation unit 107 generates at least one solution for the query received from the user. For example, for the input query "I want to get better at presentations", "To improve your presentation skills, it's important to learn the fundamental elements and techniques of presentations. For example, having a clear objective, creating content tailored to your audience, and using storytelling techniques." This generates solutions such as the following.

[0056] For example, the solution generation unit 107 generates solutions for the input text using a generative model such as a large-scale language model. When using a generative model, solutions can be generated from instruction sentences such as, "I want to improve my presentation skills, what are some solutions?"

[0057] Alternatively, for example, the solution generation unit 107 may search a large number of pre-created pairs of input queries and solution texts using the input queries and output the corresponding solutions. The search method can be the same as that used by the search unit 104 described above.

[0058] Figure 6 is a flowchart showing an example of the information processing method of the second embodiment. Step S201 is the same as step S101 (Figure 4) of the first embodiment, so its explanation is omitted.

[0059] The solution generation unit 107 generates a solution for the text combined by the joining unit 103 (step S202). Next, the search unit 104 searches for content similar to the solution generated in step S202 (step S203). Specifically, the search unit 104 converts the solution and content into vectors and calculates the similarity between the vectors. Then, the search unit 104 outputs content with a similarity equal to or greater than a predetermined threshold as search results.

[0060] Finally, the display control unit 105 displays the search results on the display device (step S204). The display control unit 105 may also generate a response statement based on the solution when displaying the search results and include it in the display information as a response statement from the system.

[0061] As described above, in the second embodiment, a solution is generated for the received query, content similar to the solution is searched for, and content similar to the solution is displayed. This allows for the recommendation of more appropriate content to the user according to the second embodiment.

[0062] (Third embodiment) Next, the third embodiment will be described. In the description of the third embodiment, explanations similar to those of the first and second embodiments will be omitted, and the differences from the first and second embodiments will be described.

[0063] In the display information of the first embodiment (Figure 3), the system's response statement is displayed. By not only displaying the content of the recommendation results but also explaining the reasons for the recommendation, user comprehension is enhanced. Therefore, in the third embodiment, the function for generating a response statement explaining the reasons for the recommendation will be described in detail.

[0064] Figure 7 shows an example of the functional configuration of the information processing device 1-3 of the third embodiment. The information processing device 1-3 of the third embodiment includes a storage control unit 100, a query receiving unit 101, a user level determination unit 102, a coupling unit 103, a search unit 104, a display control unit 105, a storage unit 106, a solution generation unit 107, and a response generation unit 108. In the third embodiment, a response generation unit 108 is further added.

[0065] In the example shown in Figure 7, the response generation unit 108 is connected by dotted lines, but depending on the information used to generate the response statement, all or some of these dotted lines may be used to connect the units.

[0066] For example, the response generation unit 108 generates a response sentence using a generative model such as a large-scale language model. Alternatively, the response generation unit 108 may output a sentence corresponding to the input information from a large number of pre-created standard sentences.

[0067] For example, the response generation unit 108 generates a response sentence using the solution generated by the solution generation unit 107. Specifically, the response generation unit 108 can use the generation model to generate a shortened version of the solution and use that as the response sentence. For example, the following sentence may be generated. "To improve your presentation skills, it's important to learn the fundamental elements and techniques of presentations. How about this?"

[0068] If the above sentence were to be generated using a generative model, it could be generated with the instruction, "Please summarize the [solution] in about 100 characters." The [solution] part would be replaced with the actual solution.

[0069] For example, the generative model can be instructed to generate a response statement using a command such as, "We will provide a solution for the request [Query]. Please explain it in about 100 characters." [Query] will contain the text generated by the join unit 103.

[0070] For example, using only a query, you can instruct the generative model to generate a response using a statement like, "Please explain the solution to the request [Query] in about 100 characters."

[0071] Furthermore, it is possible to generate a response using the solution and search results. For example, a sentence like the following can be generated. "To improve your presentation skills, it's important to learn the fundamental elements and techniques of presentations. 'Presentation Fundamentals' teaches you how to speak and create materials, while 'Improving Communication Skills' teaches you things like gestures when speaking."

[0072] If you want the above sentence to be generated by a generative model, you can use the following instruction to have the generative model generate a response: "For the request [Query], we present [Solution] and recommend [Search Result 1], [Search Result 2], and [Search Result 3]. Please explain the reasons for the recommendations." [Search Result 1], [Search Result 2], and [Search Result 3] will contain the titles and summaries of the search result content. In this example, we have shown an example for three cases, but the number of search results can be arbitrary.

[0073] The display control unit 105 displays the display information, including the response sentence such as the above sentence, on the display device.

[0074] (Fourth Embodiment) Next, the fourth embodiment will be described. In the description of the fourth embodiment, explanations similar to those of the first to third embodiments will be omitted, and the differences from the first to third embodiments will be described.

[0075] Figure 8 shows an example of the functional configuration of the information processing device 1-4 of the fourth embodiment. The information processing device 1-4 of the fourth embodiment includes a storage control unit 100, a query reception unit 101, a user level determination unit 102, a search unit 104, a display control unit 105, a storage unit 106, a solution generation unit 107, a response generation unit 108, and a solution adjustment unit 109.

[0076] The solution generation unit 107 generates at least one solution. When generating a solution using a generation model, an instruction such as "Generate a bulleted list of solutions for the request [Query]" can be used, for example, as shown below. Note that [Query] contains the text received by the query reception unit 101. " (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically "

[0077] The solution adjustment unit 109 determines an unnecessary solution from at least one solution based on the user level and the user's content learning history, and deletes the unnecessary solution.

[0078] For example, the solution adjustment unit 109 adjusts solutions using user levels. Specifically, when the solution adjustment unit 109 uses user levels based on learning history, it converts each solution and the content in the learning history into vectors and calculates the similarity between the vectors. For each solution, if there is content in the learning history with a similarity of a predetermined threshold or higher, the solution adjustment unit 109 considers this solution to be already learned and deletes it.

[0079] For example, suppose that the content shown in Figure 9 has a similarity above a certain threshold based on the learning history to the three solutions mentioned above.

[0080] Figure 9 is a diagram illustrating an example of solution adjustment 1 based on the course history of the fourth embodiment. In the example in Figure 9, the solution adjustment unit 109 has content corresponding to solutions (1) and (2), so it deletes solutions (1) and (2) and leaves only solution (3).

[0081] Furthermore, the solution adjustment unit 109 converts all content into vectors regardless of the learning history, calculates the similarity between the vectors and the solutions, and extracts content that has a similarity above a threshold for each solution. For example, if all of the extracted content is already present in the learning history, the solution adjustment unit 109 deletes that solution. Also, for example, if a certain number or percentage of the extracted content is already present in the learning history, the solution adjustment unit 109 deletes that solution.

[0082] Figure 10 is a diagram illustrating an example of adjusting solutions based on learning history in the fourth embodiment. For example, in the example in Figure 10, content with a similarity above a threshold is shown for each solution (1) to (3). Underlined content is content that is present in the learning history. Solution (1) is deleted because all similar content has been taken. Also, for example, if the setting is to delete content if more than 60% of similar content has been taken, solution (2) is deleted because 66% (=2 / 3) of it has been taken.

[0083] For example, the solution adjustment unit 109 deletes the solution if the highest-level content among the extracted content is already present in the learning history.

[0084] Figure 11 is a diagram illustrating example 3 of adjusting solutions based on the course history in the fourth embodiment. For example, in the example in Figure 11, solutions (2) and (3) are deleted because the highest level content has already been completed.

[0085] For example, the content is labeled with information about the target user, such as "for new employees" or "for managers." The solution adjustment unit 109 defines the order, such as new employees → mid-level employees → managers → executives, and deletes solutions if the target user of the extracted content is lower in rank than the actual user's career level. For example, if the user's position is "manager," solutions that only have content for new employees or mid-level employees will be deleted.

[0086] Figure 12 is a diagram illustrating an example of solution adjustment 4 based on the course history of the fourth embodiment. For example, in the example in Figure 12, if the user's job title is "Manager," then solution (1) is deleted because the similar content to solution (1) is only for new and mid-level employees.

[0087] Alternatively, the solution adjustment unit 109 may use a generation model to adjust the solution. For example, the solution adjustment unit 109 generates the following text and inputs it into the generation model. "I (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically I need to implement the following. I'm in my 40s, a manager, and work in sales. I've already taken courses on "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking (Beginner Level)," and "Improving Communication Skills." Please tell me which of the above three courses I should implement.

[0088] As described above, the system makes inquiries and prompts the generation model to respond with the number of the necessary solution based on the course history. Then, the solution adjustment unit 109 adjusts the solution based on the response from the generation model.

[0089] The response generation unit 108 generates a response based on the results of the solution adjustment unit 109. Specifically, the response generation unit 108 generates a response using at least one of the input query, solution, user level, search results, and deleted solution.

[0090] In Figure 8, the response generation unit 108 is shown connected by dotted lines, but depending on the information used to generate the response sentence, all or some of these dotted lines may be used to connect the units.

[0091] For example, the response generation unit 108 generates the following text. "To improve your presentation skills, you need to learn how to create presentation materials, how to speak effectively, and how to think logically. I believe you already have the first two skills, so I recommend learning how to think logically. Specifically, I recommend learning critical thinking and lateral thinking."

[0092] If you want to generate the above text using a generative model, "For the request [Query], we present [Solution 1], [Solution 2], and [Solution 3], and considering the [User Level], we narrow it down to [Solution n1] and [Solution n2]. Please explain your reasons for recommending them." The following instructions will generate the solutions: [Solution 1], [Solution 2], and [Solution 3] are the solutions to be generated, and the number is not limited to three. [Solution n1] and [Solution n2] are the solutions that remain after considering the user level, and the number is not limited to two.

[0093] Specifically, the instructions would look something like this: "In response to the request, 'I want to improve my presentation skills'" (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically Present, Profile: 40s, Manager, Sales Course History: "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking Skills (Beginner)," "Improving Communication Skills" Considering this, we will limit our selection to (3). Please explain your reasons for recommending it."

[0094] Alternatively, the response generation unit 108 may generate a response sentence using the search results. For example, the following sentence may be generated. "To improve your presentation skills, you need to learn how to create presentation materials, how to speak effectively, and how to think logically. Since you likely already have the first two skills, I recommend focusing on learning logical thinking. For example, how about courses like 'Beginner Critical Thinking' or 'Improving Logical Thinking Skills'?"

[0095] If the above text is to be generated by a generative model "For the request [Query], we present [Solution 1], [Solution 2], and [Solution 3]. Considering the [User Level], we limit it to [Solution n] and recommend [Search Result 1], [Search Result 2], and [Search Result 3]. Please explain the reasons for your recommendations." It can be generated using instructions like the following. [Search Result 1][Search Result 2][Search Result 3] will contain the title and summary of the search result content. Although three examples are shown in this example, the number of search results can be arbitrary.

[0096] Specifically, the instructions would look something like this: "In response to the request, 'I want to improve my presentation skills'" (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically Present, Profile: 40s, Manager, Sales Course History: "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking Skills (Beginner)," "Improving Communication Skills" Considering this, I will limit my recommendations to (3) and recommend "Critical Thinking for Beginners" and "Improving Logical Thinking Skills." Please explain your reasons for recommending them.

[0097] Figure 13 is a flowchart illustrating an example of the information processing method according to the fourth embodiment. First, the query receiving unit 101 receives text related to the user's request (step S301). Next, the solution generation unit 107 generates one or more solutions (step S302). Then, the solution adjustment unit 109 adjusts the one or more solutions generated in step S302 using the above processing (step S303).

[0098] Next, the search unit 104 searches for content similar to the solution adjusted in step S303 (step S304).

[0099] Finally, the display control unit 105 displays the search results on the display device (step S305). The display control unit 105 may also generate a response statement based on the solution when displaying the search results and include it in the display information as a response statement from the system.

[0100] (Fifth embodiment) Next, the fifth embodiment will be described. In the description of the fifth embodiment, explanations similar to those of the first to fourth embodiments will be omitted, and the differences from the first to fourth embodiments will be described. In the fifth embodiment, an embodiment in which each solution is weighted will be described.

[0101] Figures 14A and 14B show examples of the functional configurations of the information processing devices 1-5a and 1-5b of the fifth embodiment. The information processing devices 1-5a and 1-5b of the fifth embodiment include a memory control unit 100, a query reception unit 101, a user level determination unit 102, a search unit 104, a display control unit 105, a storage unit 106, a solution generation unit 107, a response generation unit 108, a weight calculation unit 110, and a weighted averaging unit 111.

[0102] The connection relationship between the weight calculation unit 110 and the weighted averaging unit 111 is different in Figures 14A and 14B.

[0103] The weight calculation unit 110 calculates the weights of the solutions. The weighted averaging unit 111 performs a weighted average of the solutions based on the vectors representing the solutions and the weights of the solutions.

[0104] For example, in the solution generation unit 107, when generating solutions using a generation model, the solutions can be prioritized by using an instruction such as "Generate solutions in bullet points in order of priority." In this case, the weight calculation unit 110 determines the weight w1 for solution s1 according to the priority based on the user level of the solutions. Then, the weighted averaging unit 111 calculates the vector v of the entire solution using the vector v1 of solution s1. If there are three solutions, the vector v of the entire solution is calculated as follows. v = w1*v1 + w2*v2 + w3*v3

[0105] When the weights calculated by the weight calculation unit 110 are used in the solution, the search unit 104 performs a search using the weighted averaged solution vector (Figure 14A).

[0106] For example, the search unit 104 performs a search using solution s1, and the weight calculation unit 110 assigns weights to the score p1 of each content in the search results. Then, the weighted averaging unit 111 calculates the final score p for each content as follows. p = w1*p1 + w2*p2 + w3*p3

[0107] When the weights calculated by the weight calculation unit 110 are used for the score p, a search is performed for each solution, and a weighted average is applied to the scores of the searched content (Figure 14B).

[0108] Since the solution generation unit 107 generates solutions in order of priority, the weights are set to the largest values ​​from top to bottom, within the range where the sum is 1. For example, w1=0.5, w2=0.3, w3=0.2. This results in a search that takes higher priority solutions into greater consideration.

[0109] Alternatively, weights can be calculated using user-level data. Specifically, weights can be determined based on the proportion of content similar to the solution that the user has not yet accessed.

[0110] Figure 15 is a diagram illustrating the user-level-based weight calculation process of the fifth embodiment. Underlined content indicates content that has already been completed.

[0111] In the example in Figure 15, the percentages of content that were not taken (k1-k3) are as follows: (1) k1 = 0 / 2 = 0 (2) k² = 2 / 3 = 0.66 (3) k3 = 2 / 2 = 1

[0112] If we adjust the above values ​​so that they sum to 1, the weights w1, w2, and w3 will be as follows. (1) w1 = k1 / (k1 + k2 + k3) = 0 (2) w2 = k2 / (k1 + k2 + k3) = 0.6 (3) w3 = k3 / (k1 + k2 + k3) = 0.4

[0113] Solution (1) is effective because all content has been completed, so w1=0, which has the same effect as deletion.

[0114] Figures 16A and 16B are flowcharts illustrating examples of information processing methods according to the fifth embodiment. The flowchart in Figure 16A shows an example of processing in the configuration of Figure 14A. The flowchart in Figure 16B shows an example of processing in the configuration of Figure 14B.

[0115] Steps S401 and S501 are the same as step S301 (Figure 13) of the fourth embodiment, so their explanation will be omitted.

[0116] After the solution generation unit 107 generates solutions (steps S402, S502), the weight calculation unit 110 calculates the weight for each solution using the solutions and the user level (steps S403, S503).

[0117] In the configuration shown in Figure 14A, as shown in Figure 16A, the weighted averaging unit 111 performs weighted averaging on the solution vector (step S404), and the search unit 104 performs content search using the weighted averaging vector (step S405).

[0118] In the configuration shown in Figure 14B, as shown in Figure 16B, the search unit 104 performs a content search for each solution (step S504), and the weighted averaging unit 111 performs a weighted average on the search scores of each content in the search results (step S505). Specifically, the weighted averaging unit 111 performs a weighted average on the search scores using the weights of the solutions used for the search.

[0119] Steps S406 and S506 are the same as step S305 (Figure 13) in the fourth embodiment. In step S506, the display control unit 105 controls the display order of the searched content based on the weighted averaged search score.

[0120] For example, the solution generation unit 107 can also determine priority using the generation model. For instance, the solution generation unit 107 generates the following text and inputs it into the generation model. "I (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically I need to do the following. I have already taken the courses "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking (Beginner)," and "Improving Communication Skills." Please tell me the priority of the above three courses on a scale of 0 to 1.

[0121] As described above, inquire and have them prioritize each solution based on your course history.

[0122] Next, the processing of the response generation unit 108 will be described. In Figure 14A, the response generation unit 108 is connected by dotted lines, but depending on the information used to generate the response statement, all or some of these dotted lines may be used to connect them.

[0123] The response generation unit 108 generates a response sentence using a generative model such as a large-scale language model. Specifically, the response generation unit 108 uses the input query, solution, and priority to generate a sentence like the one below. "To improve your presentation skills, you need to (1) learn how to create presentation materials, (2) learn speaking techniques, and (3) learn logical thinking. I believe you have already mastered (1), so I recommend that you work on (2) and (3). I especially strongly recommend (2)."

[0124] The above results are for the case where the weights are w1=0, w2=0.6, and w3=0.4.

[0125] When generating response sentences using a generative model, "For the request [Query], we present [Solution 1], [Solution 2], and [Solution 3]. Considering the [User Level], we narrow it down to [Solution n1] (Priority 1) and [Solution n2] (Priority 2). Please explain your reasons for recommending them." The following instructions will generate the solutions: [Solution 1], [Solution 2], and [Solution 3] are the solutions to be generated, and the number is not limited to three. [Solution n1] and [Solution n2] are the solutions that remain after considering the user level, and the number is not limited to two.

[0126] Specifically, the instructions would look something like this: "In response to the request, 'I want to improve my presentation skills'" (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically Present, Profile: 40s, Manager, Sales Course History: "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking Skills (Beginner)," "Improving Communication Skills" Considering this, we will limit our selection to (2) (Priority 1) and (3) (Priority 2). Please explain your reasons for recommending them."

[0127] Furthermore, in the configuration shown in Figure 14B, the search results are used "To improve your presentation skills, you need to (1) learn how to create presentation materials, (2) learn speaking techniques, and (3) learn logical thinking. I believe you have already mastered (1), so I recommend that you work on (2) and (3). In particular, I recommend that you learn 'active listening,' which is related to (2)." The following sentences are generated.

[0128] If you want to generate the above text using a generative model, "In response to the request [Query], we present [Solution 1], [Solution 2], and [Solution 3]. Considering the [User Level], we narrow it down to [Solution n1] (Priority 1) and [Solution n2] (Priority 2), and recommend [Search Result 1] (Priority n), [Search Result 2] (Priority n), and [Search Result 3] (Priority n). Please explain the reasons for your recommendations." It is generated using the following instruction.

[0129] [Search Result 1][Search Result 2][Search Result 3] will contain the title and summary of the search result content. Although three cases are shown in this example, the number of search results can be arbitrary. Since it is known which solution each piece of content was found from, the n in (priority rank n) will be replaced with the priority rank of the corresponding solution.

[0130] Specifically, the instructions would look something like this: "In response to the request, 'I want to improve my presentation skills'" (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically Present, Profile: 40s, Manager, Sales Course History: "Presentation Basics," "How to Create Presentation Materials," "Effective Speaking Skills (Beginner)," "Improving Communication Skills" Taking this into consideration, I will limit my recommendations to (2) (Priority 1) and (3) (Priority 2), and recommend "Learning Active Listening" (Priority 1), "Beginner Critical Thinking" (Priority 2), and "Improving Logical Thinking Skills" (Priority 2). Please explain your reasons for these recommendations.

[0131] (Sixth Embodiment) Next, the sixth embodiment will be described. In the description of the sixth embodiment, explanations similar to those of the first to fifth embodiments will be omitted, and the differences from the first to fifth embodiments will be described.

[0132] If any content in the search results overlaps with the user's learning history, it should be excluded. Similarly, content that is similar but not necessarily duplicates should also be excluded. Furthermore, content that is at a higher level than the recommended content should also be excluded. For example, "Basic Accounting" should not be recommended to a user who has already completed "Advanced Accounting."

[0133] Therefore, in the sixth embodiment, an embodiment for adjusting the search results will be described.

[0134] Figure 17 shows an example of the functional configuration of the information processing device 1-6 of the sixth embodiment. The information processing device 1-6 of the sixth embodiment includes a storage control unit 100, a query reception unit 101, a user level determination unit 102, a search unit 104, a display control unit 105, a storage unit 106, a solution generation unit 107, and a search result adjustment unit 112.

[0135] The search result adjustment unit 112 adjusts the search results. For example, the search result adjustment unit 112 determines unnecessary content from at least one of the searched content based on the user level and the user's content learning history, and deletes the unnecessary content.

[0136] Specifically, the search result adjustment unit 112 first removes duplicate content between the content in the learning history and the content in the search results. The search result adjustment unit 112 also performs processing to remove similar content. Specifically, the search result adjustment unit 112 converts the content in the search results and the content in the learning history into vectors and calculates the similarity between the vectors. For each search result, if there is content in the learning history with a similarity of a predetermined threshold or higher, the search result adjustment unit 112 compares the level of the content in the search results with the level of the content in the learning history. If the level of the content in the learning history is higher than the level of the content in the search results, the search result adjustment unit 112 removes the content in the search results. In other words, the search result adjustment unit 112 removes content from at least one of the searched contents that is at a level lower than the user level.

[0137] Figure 18 is a diagram illustrating Example 1 of the content adjustment process based on course history in the sixth embodiment. In the example in Figure 18, (1) is excluded from the search results because the similar course history content is of a higher level. (2) is of the same level. Whether to exclude when the levels are the same is determined in advance by a setting, and the search result adjustment unit 112 processes based on that setting. (3) is not excluded because the search result content is of a higher level.

[0138] For example, if there are multiple pieces of content at the same level, the search result adjustment unit 112 calculates the proficiency level r, and if the proficiency level r is above a predetermined threshold, it excludes the relevant search result content from the search results. First, it extracts content similar to the search result content from all content. The following is an example of extraction. The underlined similar content is content that has been taken.

[0139] Figure 19 is a diagram illustrating Example 2 of the content adjustment process based on the course history in the sixth embodiment. The underlined similar content is the content that has been taken.

[0140] First, the search result adjustment unit 112 calculates the percentage of content that users are proficient in, categorized by level. At the same level as search result content: r0 = 1 / 2 = 0.5 (Level of search result content) - 1 r1 = 2 / 2 = 1.0

[0141] The search result adjustment unit 112 calculates the proficiency level r by weighting and adding the proficiency rates for each level. Proficiency level r = w0 * r0 + w1 * r1 If we set w0=1.0 and w1=0.1, the proficiency level r becomes 0.6 as shown below. Proficiency level r = w0 * r0 + w1 * r1 = 1.0 * 0.5 + 0.1 * 1.0 = 0.6

[0142] For example, if the predetermined threshold is 0.7, this content will remain in the search results because its proficiency level r is lower than the threshold. By weighting the content in this way, the impact of content that is below the user's level can be reduced. Also, even with low-level content (e.g., level 1 content), the more the user is exposed to it, the higher their proficiency level r will become.

[0143] Furthermore, the display control unit 105 may be configured to switch between a state in which content according to the user level is not excluded and a state in which content is excluded.

[0144] Alternatively, the display control unit 105 may visually indicate whether or not content is excluded by using the content's display color and the presence or absence of a mark, rather than excluding it.

[0145] Figure 20 shows an example 1 of the display information according to the sixth embodiment. In this example, Figure 20 shows that the content grayed out by dots ("How to create presentation materials" and "Writing techniques") is excluded content.

[0146] Furthermore, the display control unit 105 may change the display order of search results according to the user level. The search unit 104 assigns a score to each piece of content. An example of a display order is "recommended order," which displays the content in descending order of score. In Figure 20, the numbers assigned to each piece of content represent the score, and the content is displayed in "recommended order."

[0147] For example, some systems prioritize displaying content that the user has not yet learned, based on their proficiency level (r).

[0148] Alternatively, the display control unit 105 may sort the content in order of increasing proficiency level r, or in order of decreasing proficiency level r. If the content is displayed in order of decreasing proficiency level r, content on topics that the user has not studied much will be displayed preferentially, thus prioritizing the display of content that fills in any knowledge gaps.

[0149] On the other hand, if content is displayed in order of proficiency level r, content related to what the user has actively studied will be displayed preferentially, thus prioritizing content that further strengthens the user's strengths.

[0150] For example, if the solution to "I want to improve my presentation skills" is as follows, (1) Learn how to create presentation materials. (2) Learning speaking techniques (3) Learning to think logically

[0151] While users can take content that covers all three solutions comprehensively, the display control unit 105 may, if the user's mastery of content related to (1) is above a predetermined threshold, display and recommend content related to (2) and (3) at a higher priority. This efficiently helps users strengthen their weaker areas of knowledge and ultimately leads to them acquiring the skills necessary for all three solutions.

[0152] Alternatively, the display control unit 105 may display the content in order of level. The display order may also be determined using other factors, such as the order in which the courses were published, the order of price, and the order of the number of participants.

[0153] Alternatively, for example, the display control unit 105 may request the user to complete a survey after they have finished viewing the content, determine the content's evaluation score based on the survey results, and display the content in order of highest evaluation score (most popular).

[0154] Figure 21 shows an example 2 of the display information in the sixth embodiment. In the example in Figure 21, the user can arbitrarily change the order of the content. Specifically, a selection box for the order is displayed so that the user can change the order, and in the example in Figure 21, "Recommended Order" is selected.

[0155] Alternatively, the display control unit 105 may integrate each order and display the content in the integrated order. For example, when the display control unit 105 integrates the recommended order s1 and the lowest price order s2, it averages the rankings determined by each order ((s1+s2) / 2) and displays the content in order of the highest average ranking. Note that the display control unit 105 may decide which order to prioritize by performing a weighted addition (w1*s1+w2*s2) rather than simply averaging. For example, by setting w1=0.7 and w2=0.3, the recommended order can be prioritized.

[0156] Integration is possible with three or more items. In Figure 21, only one order, "Recommended Order," is selected, but by selecting multiple options, the items will be sorted in the order they were integrated.

[0157] Furthermore, after the user has decided which content to take, it is also possible to determine and display the order in which the content will be taken. In the example UI (User Interface) in Figure 21, the user selects the content to take from the search results. Alternatively, if the "Select All" button is pressed, all content except for excluded content will be selected. If the "Generate Course Schedule" button is pressed after content has been selected, the content will be rearranged in the order in which it will be taken.

[0158] In addition to the order in which the content is presented as shown in Figure 21, the order in which the content is taken may also be rearranged by measuring the similarity between the content and ensuring that content with high similarity is taken consecutively. For example, the display control unit 105 may present the following order as the order in which the content is taken. "Presentation Fundamentals" → "Improving Communication Skills" → "Effective Speaking Techniques" → "Critical Thinking" → "Lateral Thinking"

[0159] "Improving communication skills" and "effective communication" are similar in content, and "critical thinking" and "lateral thinking" are similar in content, so it is more efficient to take them consecutively. The display control unit 105 suggests the most efficient order of taking the courses as described above.

[0160] The order in which courses are taken may be determined by integrating several possible sequences, similar to the display order. Furthermore, some content may have specific periods during which it can be taken. The display control unit 105 may also consider the timing of the courses, in addition to the order described above, to determine the sequence so that the courses can be completed in the shortest possible time. Alternatively, if a period is specified, the display control unit 105 may determine the sequence so that the courses can be completed within that period.

[0161] Figure 22 is a flowchart showing an example of the information processing method of the sixth embodiment. First, the query reception unit 101 receives text related to the user's request (step S601). Next, the solution generation unit 107 generates a solution (step S602). Next, the search unit 104 searches for content similar to the solution generated in step S602 (step S603).

[0162] Next, the search result adjustment unit 112 performs an adjustment process to exclude content from the search results of step S603 (step S604). Finally, the display control unit 105 displays the search results on the display device in an order based on, for example, the search result score (step S605).

[0163] (Seventh Embodiment) Next, the seventh embodiment will be described. In the description of the seventh embodiment, explanations similar to those of the first to sixth embodiments will be omitted, and the differences from the first to sixth embodiments will be described.

[0164] Figure 23 shows an example of the functional configuration of the information processing device 1-7 of the seventh embodiment. In the example in Figure 23, a response generation unit 108 is added to the configuration of the second embodiment (Figure 5), but the configuration may be based on an embodiment other than the second embodiment.

[0165] The response generation unit 108 generates a response using the search results, solutions, and input queries, but user-level information may also be used.

[0166] If the response generation unit 108 is to generate the response using a generation model, for example, it will generate the response using an instruction statement. "[User Level] For my request [Query], I propose a [Solution] and recommend [Search Result 1], [Search Result 2], and [Search Result 3]. Please explain the reason for your recommendations."

[0167] Here, [Search Result 1], [Search Result 2], and [Search Result 3] contain the title and summary of the search result content. While this example shows three results, the number of search results can be arbitrary.

[0168] The [User Level] field is where you enter text that represents the user level, as mentioned earlier. For example, the following text is possible. "I have completed courses such as 'Basic Critical Thinking,' 'Basic Accounting,' 'Intermediate Accounting,' and 'Business Etiquette.'" "I have completed the following courses: 'Basic Critical Thinking' (95 points), 'Basic Accounting' (100 points), 'Intermediate Accounting' (80 points), and 'Business Etiquette' (99 points)..." "I'm in my 40s, a section manager, in sales, and I've been with the company for 23 years." "I have a Grade Pre-2 Eiken (English Proficiency Test) and a Grade 2 Information Processing certification."

[0169] The generative model generates sentences like the one below. In the sentence below, "For those of you who have already taken the 'Business Manners' course" is the user-level-based part of the sentence. "To improve your presentation skills, it's important to learn the fundamental elements and techniques of presentations. 'Presentation Fundamentals' teaches you how to speak and create materials. If you've already taken 'Business Manners,' we recommend 'Communication Skills Improvement.' This course teaches you things like gestures when speaking."

[0170] (Eighth embodiment) Next, the eighth embodiment will be described. In the description of the eighth embodiment, explanations similar to those of the first to seventh embodiments will be omitted, and the differences from the first to seventh embodiments will be described.

[0171] Figure 24 shows an example of the functional configuration of the information processing device 1-8 of the eighth embodiment. In the example in Figure 24, a user information receiving unit 113 is added to the configuration of the second embodiment (Figure 5), but the configuration may be based on an embodiment other than the second embodiment.

[0172] The memory unit 106 maintains the conditions for users who can access each piece of content. These conditions may include, for example, the user's personal name and ID. Alternatively, the conditions may be based on organization, job title, age group, or years of service, rather than individual users. Specifically, conditions such as being in the Second Sales Department, at least a manager, 40 years of age or older, and having 10 or more years of service may be specified. Multiple conditions may be combined. Content is only provided to users who meet the specified conditions.

[0173] The user information reception unit 113 receives user information of users accessing the content. This user information is used to determine whether or not the user is allowed to access the content. For example, the user information reception unit 113 receives personal name, ID, organization, job title, age, and years of employment.

[0174] The display control unit 105 uses the search results and user information to control the content displayed according to the display information. For example, the display control unit 105 removes content that the user cannot access from at least one searched content item based on the user information. Alternatively, the display control unit 105 distinguishes content that the user cannot access from at least one searched content item based on the user information and displays the display information accordingly.

[0175] Figure 25 shows an example of display information 1 of the eighth embodiment. In the example in Figure 25, the third and fourth items from the top of the searched content are inaccessible content. Inaccessible content may be displayed in black as shown in Figure 25, or it may be excluded from the search results.

[0176] Figure 26 shows an example 2 of the display information according to the eighth embodiment. In the example in Figure 26, inaccessible content may be displayed only as the content title and score, and may be shown in a way that prevents selection.

[0177] (Ninth Embodiment) Next, the ninth embodiment will be described. In the description of the ninth embodiment, explanations similar to those of the first to eighth embodiments will be omitted, and the differences from the first to eighth embodiments will be described.

[0178] Figures 27A to 27C show examples of the functional configurations of the information processing devices 1-9a to 1-9c of the ninth embodiment. In the examples in Figures 27A to 27C, a response generation unit 108 and a system action determination unit 114 are added to the configuration of the second embodiment (Figure 5), but the configuration may be based on an embodiment other than the second embodiment.

[0179] The system action determination unit 114 determines a system action for the query (first query) received from the user. The response generation unit 108 generates a response statement corresponding to the system action.

[0180] Examples of system actions are as follows: "Display Search Results": Generates solutions, performs searches, displays search results, and generates response text. "Solution Selection": Generates solutions and allows the user to select the appropriate solution. "Confirmation": Check if this is the correct solution. "Refinement Request": The query is ambiguous, so we request input that will help refine it. "Re-entry Request": The query was not understood, so please re-enter the information.

[0181] "Displaying search results" is a system action that performs the processing described in the above embodiment.

[0182] "Solution Selection" is a system action that prompts the system to select the appropriate solution when multiple solutions are generated. For example, suppose the following four solutions are generated. 1. Read books and textbooks on business administration: Reading books and textbooks on business administration is beneficial for learning the basic concepts and theories of business administration. 2. Participate in online courses or business schools: You can deepen your expertise by participating in online business administration courses or business schools. 3. Gain practical experience: Gaining experience in real business settings is also important when studying business administration. You can acquire practical management skills by starting your own business or through internships and volunteer activities. 4. Talk to business owners and business consultants: By talking to experts such as business owners and business consultants, you can gain knowledge and experience about actual business management. It is recommended to participate in business events and seminars, or to schedule meetings with experts.

[0183] The system action determination unit 114 instructs the response generation unit 108 to generate a response statement that allows the user to select one or more solutions from these four options. For example, the response generation unit 108 might generate a statement such as, "Please select the solution you feel is necessary from these options."

[0184] "Confirmation" is a system action to verify that the generated solution is acceptable. For example, the response generation unit 108 generates a response message such as, "A solution like ~ has been generated, is this OK?"

[0185] A "refinement request" is a system action that prompts the user to enter additional queries if the received query is insufficient to generate a solution. For example, the following cases are examples of system actions that would trigger a "refinement request." User: "I want to become a respected manager." System: "To become a respected manager, you can develop leadership skills, improve your communication abilities, and build a team. What skills would you like to develop?"

[0186] A "re-entry request" is a system action that prompts a user to re-enter their input query when the system cannot understand it. For example, the following cases are examples of system actions that would trigger a "re-entry request." User: "I wonder if it will be sunny tomorrow." System: "I'm sorry, I didn't understand. Could you please repeat that?"

[0187] In the configuration shown in Figure 27A, the system action determination unit 114 determines which action to output based on the received query. However, in the configuration shown in Figure 27A, the system action is generated before the solution is generated, so the system action is selected from three options: "Display search results," "Request filtering," and "Request re-entry."

[0188] In the configuration shown in Figure 27B, the system action determination unit 114 determines which action to output based on the received query and solution.

[0189] In the configuration shown in Figure 27C, the system action determination unit 114 determines which action to output based on the received query and the response statement generated by the response generation unit 108. The system action determination unit 114 may also determine the system action based on the search results output by the search unit 104.

[0190] For example, the system action decision unit 114 may be implemented by a rule-based decision mechanism. Alternatively, the system action decision unit 114 may be implemented by a machine learning model that has been trained on a large amount of data in advance.

[0191] For example, the system action determination unit 114 may be implemented by a large-scale language model. When using a large-scale language model to determine a system action, the following instruction statements are possible. " In response to the statement, "I want to become a respected manager," you choose one of the following five actions. Please rate the appropriateness of each action on a scale of 1 to 10. Search results displayed: "Here are some solutions for becoming a respected manager... We recommend this content." Solution Selection: "Here are some solutions for becoming a respected manager... Please choose the solution you feel is most necessary." Confirmation: "Here are some solutions for becoming a respected manager... Is this alright?" Refinement Request: "To become a respected manager, you can develop leadership skills, improve communication skills, and implement team-building strategies. What skills would you like to develop?" Request for re-entry: "I'm sorry, I didn't understand. Could you please repeat that?" "

[0192] The large-scale language model assigns scores to the five actions described above. The system action determination unit 114 selects the action with the highest score. Alternatively, instead of assigning scores, the system action determination unit 114 may have the large-scale language model rank the actions.

[0193] If a system action such as "refinement request" or "confirmation" is selected, the next query accepted will be a continuation of the previous query. Therefore, the system action determination unit 114 needs to consider not only the immediately preceding query, but also the previous query and the system response statement. A series of inputs that require consideration of previous queries in this way is called an interactive session.

[0194] When "Display Search Results" is selected, the conversation session ends. Subsequent queries will be treated as separate conversation sessions. Similarly, when "Request Re-entry" is selected, the conversation session also ends.

[0195] During the ongoing dialogue session, past inputs are taken into consideration when determining system actions, generating solutions, and producing responses. Specifically, data combining previously received queries and system response statements is used as input.

[0196] For example, the instruction statement for implementing the solution generation unit 107 using a generation model would be as follows: "You (the system) had the following conversation with the user. What solution would you recommend to the user?" User: "I want to become a respected manager." System: "To become a respected manager, you can develop leadership skills, improve your communication abilities, and build a team. What skills would you like to develop?" User: "I want to demonstrate leadership." "

[0197] Furthermore, the information processing devices 1 to 1-9 of the first to ninth embodiments described above are applicable to content other than educational content. For example, products on an e-commerce site are accompanied by information such as product name, overview, and specifications, and since they contain text information, they can be searched. For example, in the case of sports equipment, it is possible to recommend appropriate products by considering the user's skill level, physique, and age as levels.

[0198] For example, in response to the query, "I want a sports bike that can easily handle mountain trails," the following solutions can be offered. When choosing a sports bike, you should consider the following factors. Frame material: Lightweight yet sturdy aluminum or carbon fiber are recommended. Number of gears: More gears allow for smoother riding on mountain roads and uphill slopes. Generally, 20 speeds or more are recommended. Suspension: To make riding on mountain roads more comfortable, it's a good idea to choose a motorcycle with suspension on the front forks. Brakes: Reliable disc brakes are recommended to ensure sufficient stopping power on mountain roads. Tires: Choosing high-grip mountain bike tires will allow for stable riding on mountain trails. Taking these factors into consideration, it's a good idea to choose a sports bike that suits your budget and preferences. It's also recommended to actually test ride bikes at specialty shops or bicycle retailers. "

[0199] In this way, vague requests can be translated into specific specifications. Furthermore, by recommending content tailored to the user's physique, users can select content that meets their needs without needing to know the detailed specifications of sports bikes.

[0200] Finally, we will describe examples of the hardware configurations of the information processing devices 1 to 1-9 of the first to ninth embodiments.

[0201] [Example hardware configuration] Figure 28 shows an example of the device configuration of the information processing devices 1 to 1-9 of the first to ninth embodiments. The information processing devices 1 to 1-9 of the first to ninth embodiments include a processor 201, a main memory 202, an auxiliary storage device 203, a display device 204, an input device 205, and a communication device 206. The processor 201, main memory 202, auxiliary storage device 203, display device 204, input device 205, and communication device 206 are connected via a bus 210.

[0202] Note that information processing devices 1 to 1-9 may not be equipped with some of the above configurations. For example, if information processing devices 1 to 1-9 can utilize the input and display functions of an external device, then information processing devices 1 to 1-9 may not be equipped with the display device 204 and the input device 205.

[0203] The processor 201 executes the program read from the auxiliary storage device 203 into the main memory device 202. The main memory device 202 is memory such as ROM and RAM. The auxiliary storage device 203 is such as an HDD (Hard Disk Drive) and a memory card.

[0204] The display device 204 is, for example, a liquid crystal display. The input device 205 is an interface for operating the information processing devices 1 to 1-9. The display device 204 and the input device 205 may be implemented by a touch panel or the like that has both display and input functions. The communication device 206 is an interface for communicating with other devices.

[0205] For example, programs executed by information processing devices 1 to 1-9 are provided as computer program products, recorded in installable or executable file format on computer-readable storage media such as memory cards, hard disks, CD-RWs, CD-ROMs, CD-Rs, DVD-RAMs, and DVD-Rs.

[0206] Alternatively, for example, the programs executed by the information processing devices 1 to 1-9 may be stored on a computer connected to a network such as the Internet, and provided by being downloaded via the network.

[0207] Alternatively, for example, the information processing devices 1 to 1-9 may be configured to provide programs via a network such as the Internet without requiring downloads. Specifically, the information processing may be executed by a so-called ASP (Application Service Provider) type service, where the server computer does not transfer programs, but instead implements the processing function only by issuing execution instructions and obtaining results.

[0208] Alternatively, for example, the programs for information processing devices 1 to 1-9 may be pre-installed and provided in ROM or the like.

[0209] The programs executed by the information processing devices 1 to 1-9 are configured as modules that include functions that can also be implemented by programs, as described above. In actual hardware terms, each of these functions is loaded onto the main memory 202 by the processor 201 reading and executing a program from the storage medium. In other words, each of these function blocks is generated on the main memory 202.

[0210] Furthermore, some or all of the above-mentioned functions may be implemented using hardware such as an IC (Integrated Circuit) instead of software.

[0211] Alternatively, multiple processors 201 may be used to implement each function, in which case each processor 201 may implement one of the functions, or two or more of the functions.

[0212] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims of the invention and its equivalents. [Explanation of symbols]

[0213] 1. Information Processing Device 100 Memory Control Unit 101 Query Reception Department 102 User Level Determination Unit 103 Joint 104 Search Section 105 Display Control Unit 106 Storage section 107 Solution generation part 108 Response generation unit 109 Solution Coordination Department 110 Weight calculation unit 111 Weighted Average Section 112 Search Results Adjustment Department 113 User Information Reception Department 114 System Action Determination Unit 201 Processor 202 Main storage 203 Auxiliary storage device 204 Display device 205 Input device 206 Communication equipment 210 Bus

Claims

1. A computer having a storage device that stores content and the level of said content, A query receiving unit that receives a first query used to search for the aforementioned content, A user level determination unit that determines a user level indicating the level of at least one skill of the user, A search unit that searches for the content based on the first query and the user level, A display control unit that displays display information including at least one of the retrieved contents on a display device. A program designed to function as such.

2. The aforementioned computer, By combining the sentence indicating the user level and the sentence indicating the first query, it further functions as a coupling unit that generates a second query. The search unit searches for the content using the second query. The program according to claim 1.

3. The aforementioned computer, It further functions as a solution generation unit that generates at least one solution based on the first query, The search unit searches for the content using the solution. The program according to claim 1.

4. The aforementioned computer, A coupling unit that generates a second query by combining the sentence indicating the user level and the sentence indicating the first query, It further functions as a solution generation unit that generates at least one solution based on the second query, The search unit searches for the content using the solution. The program according to claim 1.

5. The aforementioned computer, It further functions as a response generation unit that generates a response statement corresponding to the aforementioned solution, The display control unit displays the display information, which further includes the response statement, on the display device. The program according to claim 3 or 4.

6. The aforementioned computer, It further functions as a solution adjustment unit that adjusts at least one of the solutions using the user level, The search unit searches for the content using the adjusted solution. The program according to claim 3 or 4.

7. The solution adjustment unit determines an unnecessary solution from at least one of the solutions based on the user level and the user's content learning history, and deletes the unnecessary solution. The program according to claim 6.

8. The aforementioned computer, A weight calculation unit calculates the weight of the solution according to the priority of the solution based on the user level, The weights of the aforementioned solutions are used to further function as a weighted averaging unit that performs a weighted averaging of the aforementioned solutions. The search unit searches for the content using the weighted averaged solution. The program according to claim 3 or 4.

9. The aforementioned computer, It functions as a weight calculation unit that calculates the weight of the solution according to the priority of the solution based on the user level, The search unit searches the content using the solution and outputs search results and a search score. The aforementioned search score is further configured to function as a weighted averaging unit, which performs a weighted averaging using the weights of the solutions used in the search. The display control unit controls the display order of the searched content based on the weighted averaged search score. The program according to claim 3 or 4.

10. The aforementioned computer, A search result adjustment unit determines unnecessary content from at least one of the searched contents based on the user level and the user's content learning history, and deletes the unnecessary content. The program according to any one of claims 1 to 4, which further functions as such.

11. The search result adjustment unit removes content from at least one of the searched contents that is at a level lower than the user level. The program according to claim 10.

12. The aforementioned computer, A response generation unit that generates a response statement using the first query, the user level, and at least one of the retrieved content, The program according to any one of claims 1 to 4, which further functions as such.

13. The aforementioned computer, In response to the aforementioned first query, it further functions as a system action determination unit that determines a system action, The response generation unit generates a response statement corresponding to the system action. The program according to claim 12.

14. The aforementioned computer, It further functions as a user information receiving unit that receives user information of users accessing the aforementioned content, The display control unit, based on the user information, deletes content that the user cannot access from at least one of the retrieved contents, or distinguishes content that the user cannot access, and displays the display information. The program according to any one of claims 1 to 4.

15. A storage device that stores the content and the level of the content, A query receiving unit that receives a first query used to search for the aforementioned content, A user level determination unit that determines a user level indicating the level of at least one skill of the user, A search unit that searches for the content based on the first query and the user level, A display control unit that displays display information including at least one of the retrieved contents on a display device, An information processing device equipped with the following features.

16. An information processing method for an information processing apparatus comprising a storage device that stores content and the level of said content, The information processing device receives a first query used to search for the content, The information processing device includes the steps of determining a user level that indicates the level of at least one skill of the user, The information processing device performs the steps of searching for the content based on the first query and the user level, The information processing device includes the step of displaying display information containing at least one retrieved piece of content on a display device, Information processing methods including

Images