Input generating device, summary generating system, and program

Abstract

To generate appropriate summary information.SOLUTION: An input generating device for generating input information to be input to a processing model so that the processing model generates summary information of a plurality of pieces of candidate information includes: a classification part which classifies the plurality of pieces of candidate information into one or more pieces of selection information selected by a user and one or more pieces of non-selection information not selected by the user; and an input generating device part which generates input information according to a content of the one or more pieces of selection information and a content of the one or more pieces of non-selection information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an input generation device, a summary generation system, and a program. [Background technology]

[0002] Patent Documents 1 to 3 disclose techniques for creating summaries of multiple documents. Patent Document 1: Patent No. 4938298 Patent Document 2: JP 2022-25665 A Patent Document 3: JP 2023-7372 A Summary of the Invention

[0003] A first aspect of the present invention provides an input generation device that generates input information to be input to a processing model, causing the processing model to generate summary information of a plurality of pieces of candidate information. The input generation device may include a classification unit that classifies the plurality of pieces of candidate information into one or more pieces of selected information selected by a user and one or more pieces of non-selected information not selected by the user. The input generation device may include an input generation unit that generates the input information based on the content of one or more pieces of selected information and the content of one or more pieces of non-selected information.

[0004] In any of the above input generation devices, the input generation unit may generate one or more pieces of exclusion information indicating items included in the non-selected information, and generate the input information that generates the summary information by excluding the exclusion information from the contents of the selected information.

[0005] In any of the above input generation devices, the input generation section may extract at least some of the words included in the non-selected information as the excluded information.

[0006] Any of the above input generation devices may include an information collection unit that collects the plurality of pieces of candidate information based on collection conditions input by the user. Any of the above input generation devices may include a presentation unit that presents the collected plurality of pieces of candidate information to the user. In any of the above input generation devices, the input generation unit may generate the exclusion information based on the collection conditions.

[0007] In any of the above input generation devices, the input generation unit may present the exclusion information generated from the non-selected information to the user, obtain instructions from the user regarding the exclusion information, and determine the exclusion information to be included in the input information.

[0008] Any of the input generation devices may include a history storage unit that records the user's past instructions. In any of the input generation devices, the input generation unit may generate the exclusion information based on the same user's past instructions.

[0009] In any of the above input generation devices, the input generation unit may calculate a similarity between each piece of excluded information and the selected information, and present one or more pieces of excluded information arranged based on the calculated similarity to the user.

[0010] In any of the above input generation devices, the input generation unit may calculate a similarity between each piece of exclusion information and the selected information, and determine the exclusion information to be included in the input information based on the calculated similarity.

[0011] In any of the above input generation devices, the input generation unit may calculate a similarity between each piece of non-selected information and the selected information, and generate the exclusion information corresponding to each piece of non-selected information based on the calculated similarity.

[0012] In any of the above input generation devices, the input generation unit may acquire timing information regarding the creation time of each piece of non-selected information, and generate the exclusion information corresponding to each piece of non-selected information based on the timing information.

[0013] In any of the above input generation devices, the input generation unit may present the exclusion information and one or more provisional summaries generated without using the exclusion information based on the selection information to the user, and generate the input information based on instructions obtained from the user regarding the exclusion information.

[0014] In a second aspect of the present invention, there is provided a summary generation system for generating summary information of a plurality of pieces of candidate information. The summary generation system may include a classification unit that classifies the plurality of pieces of candidate information into one or more pieces of selected information selected by a user and one or more pieces of non-selected information not selected by the user. The summary generation system may also include a summary generation unit that generates the summary information based on the content of one or more pieces of selected information and the content of one or more pieces of non-selected information.

[0015] In a third aspect of the present invention, there is provided a program for causing a computer to function as the input generation device of the first aspect.

[0016] The above summary of the invention does not list all of the features of the present invention, and subcombinations of these features may also constitute inventions. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a block diagram illustrating an example of a summary generation system 10 according to an embodiment of the present invention. [Figure 2] 10 is a diagram showing an example of a plurality of pieces of candidate information presented by a presentation unit 102. FIG. [Figure 3] FIG. 10 is a diagram illustrating an example of the operation of the classification unit 110. [Figure 4] 10 is a diagram showing an example of input information generated by an input generating unit 120. FIG. [Figure 5] 10 is a diagram showing an example of summary information acquired by a summary acquisition unit 130. FIG. [Figure 6] 10A and 10B are diagrams illustrating another example of the operation of the input generation device 100. [Figure 7] 3 is a diagram showing an example of second input information generated by the input generation device 100. FIG. [Figure 8] 10 is a diagram showing an example of corrected summary information acquired by the summary acquisition section 130. FIG. [Figure 9] FIG. 10 is a diagram illustrating another exemplary configuration of the input generation device 100. [Figure 10] 10 is a diagram illustrating an example of the operation of the exclusion information editing unit 150. FIG. [Figure 11] FIG. 10 is a diagram showing another example of exclusion information presented to the user. [Figure 12] FIG. 10 is a diagram illustrating another example of a method for generating exclusion information. [Figure 13] FIG. 10 is a diagram illustrating another exemplary configuration of the input generation device 100. [Figure 14] 10 is a flowchart illustrating an example of the operation of the input generation device 100. [Figure 15] 10 is a diagram illustrating another example of the operation of the input generation device 100. FIG. [Figure 16] 10 is a flowchart illustrating an example of the operation of the input generation device 100. [Figure 17] 12 illustrates an example computer 1200 in which aspects of the present invention may be embodied, in whole or in part. DETAILED DESCRIPTION OF THE INVENTION

[0018] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention according to the claims. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0019] FIG. 1 is a block diagram showing an example of a summary generation system 10 according to an embodiment of the present invention. The summary generation system 10 generates summary information for a plurality of pieces of candidate information. The candidate information may include document data or image data. The summary information may include a summary document that summarizes the document of the candidate information, an explanatory document that explains the content of the image of the candidate information, or both a summary document and an explanatory document.

[0020] The summary generation system 10 includes an input generation device 100 and a summary generation unit 200. The summary generation unit 200 has a processing model 210 that operates in response to input information. The summary generation unit 200 may include one or more information processing devices that operate the processing model 210. In the summary generation unit 200, the multiple information processing devices may cooperate to realize processing by the processing model 210.

[0021] The processing model 210 may be generated by machine learning using training data. The processing model 210 may use a publicly available general-purpose generative artificial intelligence (also referred to as generative AI). The generative artificial intelligence in this example is generated by deep learning. The generative artificial intelligence is capable of generating outputs not only for a certain type of explanatory variables but also for a variety of requests or questions. The processing model may be a large-scale language model. The processing model 210 may be a GPT (registered trademark) (Generative Pre-trained Transformer), a BERT (Bidirectional Encoder Representations from Transformers), or another language model.

[0022] The input generation device 100 generates input information to be input to the processing model 210 so that the processing model 210 generates summary information of a plurality of pieces of candidate information. The input information includes at least a portion of the content of the candidate information to be summarized. The input information may include instructions for causing the processing model 210 to generate summary information of the candidate information. The input information may be a so-called prompt input to the generation AI.

[0023] The input generation device 100 includes one or more information processing devices. Multiple information processing devices may work together to function as the input generation device 100. The information processing devices function as each component of the input generation device 100. Each information processing device may be a computer terminal managed by a user receiving a service, a server managed by a service provider, or another computer. Each information processing device may be connected by a general-purpose network such as the Internet, or may be connected by a dedicated line. Each information processing device may be connected by a wireless line or a wired line.

[0024] The input generation device 100 includes a classification unit 110 and an input generation unit 120. The input generation device 100 may further include a presentation unit 102 and a summary acquisition unit 130. The presentation unit 102 presents a plurality of pieces of candidate information to a user. The user is a recipient of a service who is to receive summary information of the plurality of candidate information. The presentation unit 102 may include a display device that displays the plurality of pieces of candidate information to the user, and may also include a communication device that transmits the plurality of candidate information to the user's terminal.

[0025] The user selects one or more pieces of candidate information from among the plurality of pieces of candidate information. In this example, the user selects the candidate information to be reflected in the summary information. The presentation unit 102 may present the plurality of pieces of candidate information to the user in a manner that allows the user to select the candidate information to be reflected in the summary information. For example, the presentation unit 102 may present an input section, such as a checkbox, associated with each piece of candidate information, allowing the user to input whether or not to select the candidate information.

[0026] The classification unit 110 classifies the plurality of pieces of candidate information into one or more pieces of selected information selected by the user and one or more pieces of non-selected information not selected by the user. The classification unit 110 may classify the candidate information selected by the user through the above-described input portion as selected information, and may classify the candidate information not selected by the user as non-selected information.

[0027] The input generation unit 120 generates input information based on the contents of one or more pieces of selected information and the contents of one or more pieces of non-selected information. The input generation unit 120 may generate input information that includes all of the contents of each piece of selected information, or may generate input information that includes at least a portion of the contents of each piece of selected information. The input generation unit 120 may generate input information that includes all of the contents of each piece of non-selected information, or may generate input information that includes at least a portion of the contents of each piece of non-selected information.

[0028] The input generation unit 120 may generate a single piece of input information or multiple pieces of input information based on the selected information and the non-selected information. Any piece of input information may be generated using only the selected information out of the selected information and the non-selected information. Any piece of input information may be generated using only the non-selected information out of the selected information and the non-selected information. As described above, in this specification, even if any piece of input information is generated using only the selected information and any piece of input information is generated using only the non-selected information, this is interpreted as the input information being generated based on the selected information and the non-selected information. Multiple pieces of input information may be input to the processing model 210 simultaneously or sequentially. The input generation unit 120 may generate the next piece of input information depending on the output of the processing model 210 for any piece of input information.

[0029] The input generating unit 120 may include in the input information the content of the non-selected information together with information indicating that the non-selected information was not selected by the user. The input information may include an instruction to generate a summary of the multiple selected information items in consideration of the fact that the content of the non-selected information was not selected by the user.

[0030] The processing model 210 generates summary information based on the input information. In this example, the input information includes not only the selection information but also information about what content the user did not select. This allows the generated summary information to better reflect the user's intentions compared to when summary information is generated from the selection information alone.

[0031] The input generation unit 120 may generate one or more pieces of exclusion information indicating matters included in the non-selected information, and may generate input information including an instruction to generate summary information by excluding the exclusion information from the content of the selected information. The exclusion information may be a word, a phrase, or a single sentence included in the non-selected information. The exclusion information is not limited to content directly excerpted from the non-selected information. The exclusion information may be a summary of the non-selected information, a main topic of the non-selected information, a problem to be solved by technical content described in the text of the non-selected information, or an effect achieved by the technical content. The exclusion information may be a superordinate or subordinate concept of the content of the non-selected information. The exclusion information may include multiple items, such as multiple words. In this specification, each item may be treated as one piece of exclusion information. For example, multiple words may be treated as multiple pieces of exclusion information. By generating the exclusion information, the input generation unit 120 can generate clearer instructions for the processing model 210 and obtain more appropriate summary information.

[0032] The input generation unit 120 may generate exclusion information based on information about each user, such as the user's selection history of candidate information. This allows summary information appropriate for each user to be obtained. When the summary generation unit 200 receives input information from multiple input generation devices 100, each input generation device 100 manages information about the user, so that the summary generation unit 200 does not need to manage information about all users. This reduces the storage capacity for user information in the summary generation unit 200 and also reduces the amount of data processing in the summary generation unit 200 for generating exclusion information.

[0033] The summary acquisition unit 130 acquires the summary information generated by the processing model 210. The summary acquisition unit 130 may provide the summary information to a user. The summary acquisition unit 130 may have a display device that displays the summary information to the user, and may also have a communication device that transmits the summary information to the user's terminal.

[0034] 2 is a diagram showing an example of multiple pieces of candidate information presented by the presentation unit 102. In this example, the candidate information includes text data. The presentation unit 102 may present an identification number that identifies each piece of candidate information and an input section 104 that accepts an input as to whether or not to select each piece of candidate information, in association with each piece of candidate information.

[0035] The user switches whether or not to select each piece of candidate information by operating the input section 104. In the example of Fig. 2, candidate information with a black check box in the input section 104 has been selected by the user, and candidate information with a white check box in the input section 104 has not been selected by the user. In the example of Fig. 2, candidate information Nos. 3, 5, and 6 have been selected by the user, and candidate information Nos. 1, 2, and 4 have not been selected by the user.

[0036] 3 is a diagram showing an example of the operation of the classification unit 110. As described above, the classification unit 110 classifies each piece of candidate information into selected information and non-selected information according to the selection result by the user. In the example of FIG. 3, the candidate information Nos. 3, 5, and 6 are classified as selected information, and the candidate information Nos. 1, 2, and 4 are classified as non-selected information.

[0037] The classification unit 110 in this example generates exclusion information based on the non-selected information. The exclusion information in this example is obtained by extracting at least some of the words included in the non-selected information. The classification unit 110 may extract at least one word from each piece of non-selected information to set as exclusion information. The classification unit 110 may also extract words or synonyms commonly included in two or more pieces of non-selected information to set as exclusion information. In the example of FIG. 3, the classification unit 110 extracts the words "vitamin C" and "collagen production," which are commonly included in the non-selected information items No. 1 and No. 4, as exclusion information. The classification unit 110 also extracts "skin" and synonyms of "skin," which are commonly included in the non-selected information items No. 1, No. 2, and No. 4, as exclusion information.

[0038] FIG. 4 is a diagram showing an example of input information generated by the input generation unit 120. The input information in this example includes first instruction information, selection information, second instruction information, and exclusion information. The first instruction information and second instruction information are information that instructs the content of the processing to be performed by the processing model 210. The first instruction information in this example is an instruction to generate summary information of the selection information, and the second instruction information is an instruction to not include the exclusion information in the summary information. The content of each piece of instruction information may be set in advance in the input generation unit 120.

[0039] The selection information in this example includes all of the contents of the candidate information selected by the user. The exclusion information in this example is the same as the example in FIG.

[0040] FIG. 5 is a diagram showing an example of summary information acquired by the summary acquisition unit 130. The summary acquisition unit 130 acquires summary information generated by the processing model 210 in response to input information. As shown in FIGS. 4 and 5, the summary information summarizes the contents of one or more pieces of selected information and does not include the contents indicated in the excluded information. In this way, by generating a summary of the selected information using the contents of the non-selected information, it is possible to generate summary information that better reflects the user's intentions.

[0041] The processing model 210 may generate summary information for each selected piece of information so that at least some of the content is included. In other words, the processing model 210 may generate summary information so as not to exclude all of the content of any of the selected information.

[0042] 6 is a diagram illustrating another example of the operation of the input generation device 100. In the examples of FIGS. 2 to 5, the input generation device 100 generates one piece of input information from selected information and non-selected information. In this example, the input generation device 100 divides the content of the input information as shown in FIG. 4 into first input information and second input information, and sequentially inputs them to the processing model 210. Even with this processing, it is possible to obtain summary information similar to the examples of FIGS. 2 to 5.

[0043] The input generation device 100 of this example generates first input information based on the selection information. The first input information of this example includes information similar to the first instruction information and selection information in the input information shown in Figure 4. The input generation device 100 inputs the first input information to the processing model 210 to generate a provisional summary. The provisional summary may also include exclusion information for non-selected information.

[0044] 7 is a diagram showing an example of second input information generated by the input generation device 100. The second input information is an instruction for obtaining revised summary information by excluding the content of non-selected information from the provisional summary obtained in response to the first input information. The input generation device 100 in this example generates the second input information based on the provisional summary and non-selected information. The second input information may include second instruction information and exclusion information similar to the input information shown in FIG. 4, as well as a provisional summary.

[0045] 8 is a diagram showing an example of corrected summary information acquired by the summary acquisition unit 130. The summary acquisition unit 130 acquires the corrected summary information generated by the processing model 210 in response to the second input information. Even if the input information is divided into multiple prompts as in this example, summary information that better reflects the user's intentions can be generated, as in the examples of FIGS. 2 to 5.

[0046] The input generator 120 may present one or more preliminary summaries and exclusion information to the user. As described above, the preliminary summaries are summaries generated by the processing model 210 based on the selection information and without using the exclusion information. The input generator 120 may generate the input information based on instructions obtained from the user regarding the exclusion information. The input generator 120 may generate the input information by reflecting user modifications to the exclusion information. User modifications to the exclusion information are similar to the modification instructions described below.

[0047] 9 is a diagram showing another example of the configuration of the input generation device 100. The input generation device 100 of this example includes an exclusion information editing unit 150 and a history storage unit 160. The configuration other than the exclusion information editing unit 150 and the history storage unit 160 is the same as that of the input generation device 100 of any of the aspects described in this specification.

[0048] The input generation unit 120 of this example presents the exclusion information generated from the non-selected information to the user via the exclusion information editing unit 150. The input generation unit 120 acquires correction instructions from the user regarding the exclusion information via the exclusion information editing unit 150. The correction instructions may include the result of deleting or selecting part of the presented exclusion information, or the result of adding or changing content to the presented exclusion information. The input generation unit 120 determines the exclusion information to be included in the input information based on the correction instructions from the user. The input generation unit 120 may determine the exclusion information to be included in the input information by reflecting some or all of the correction instructions from the user in the presented exclusion information. The exclusion information editing unit 150 may include a display device that displays the exclusion information, an input means for inputting correction instructions from the user, and a communication device for transmitting and receiving the exclusion information and correction instructions to and from the user's terminal.

[0049] The history storage unit 160 records past correction instructions from users for the exclusion information. The history storage unit 160 may be provided for each input generation device 100. The history storage unit 160 may also be provided in common for a plurality of input generation devices 100. In this case, even when users use different input generation devices 100, the history for each user can be consolidated and managed.

[0050] The input generation unit 120 may automatically generate exclusion information based on past correction instructions of the same user. For example, the input generation unit 120 may extract correction instructions that meet a predetermined condition from past correction instructions of the same user and reflect them in the exclusion information. The condition may specify a period during which the correction instructions were generated. For example, the input generation unit 120 may reflect correction instructions generated within a recent predetermined period in the exclusion information. The condition may be generated from the exclusion information presented to the user. For example, the input generation unit 120 may extract correction instructions for past exclusion information that have at least a portion in common with the content included in the exclusion information to be edited, and reflect them in the exclusion information to be edited.

[0051] FIG. 10 is a diagram illustrating an example of the operation of the exclusion information editing unit 150. As described above, the exclusion information editing unit 150 presents the exclusion information to be edited to the user. The exclusion information editing unit 150 acquires a correction instruction from the user for the exclusion information to be edited. In this example, the correction instruction is an instruction to delete some of the multiple words included in the exclusion information. The input generation unit 120 generates input information for the processing model 210 using the exclusion information that reflects the correction instruction from the user.

[0052] According to this example, it is possible to generate exclusion information that more accurately reflects the user's intentions. As a result, more appropriate summary information can be generated. Furthermore, by correcting the exclusion information using the user's past correction instructions, it is possible to generate exclusion information that reflects the user's correction trends. For example, words deleted by the user from the exclusion information are likely to be words that the user is interested in. For example, by deleting words deleted from multiple pieces of past exclusion information from the exclusion information to be edited, it is possible to generate exclusion information that is in line with the user's tendencies, and more appropriate summary information can be obtained.

[0053] 11 is a diagram showing another example of exclusion information presented to the user. In this example, the input generation unit 120 calculates the similarity between each piece of exclusion information to be edited and the selected information via the exclusion information editing unit 150, and presents one or more pieces of exclusion information arranged based on the calculated similarity to the user.

[0054] In this example, each piece of excluded information is a word included in the non-selected information. The input generation unit 120 calculates the similarity between each word and the selected information. The similarity may be calculated based on whether or not each word or synonym is included in any of the selected information. The input generation unit 120 may calculate a higher similarity as the number of selected information pieces that include the word or synonym increases. However, the method for calculating the similarity is not limited to this. The input generation unit 120 may calculate the similarity using a known method, such as vectorizing a sentence and calculating the similarity.

[0055] The input generation unit 120 may present the exclusion information to the user in descending or ascending order of the calculated similarity. The input generation unit 120 may also present the similarity of each piece of exclusion information to the user in association with each piece of exclusion information. The user excludes information that the user does not want excluded from the summary from the exclusion information. This can assist the user in excluding information.

[0056] When the number of pieces of excluded information reaches or exceeds a predetermined reference value, the input generating unit 120 may associate the pieces of excluded information with the similarity and present them to the user. This prevents too much information from being excluded from the summary information.

[0057] In another example, the input generation unit 120 may automatically determine the exclusion information to be included in the input information based on the similarity of each piece of exclusion information. That is, the input generation unit 120 may determine the exclusion information without relying on a correction instruction from the user. For example, when the number of pieces of exclusion information generated from non-selected information is equal to or greater than a predetermined reference value, the input generation unit 120 may delete the exclusion information in descending order of similarity until the number of pieces of exclusion information falls below the reference value.

[0058] FIG. 12 is a diagram showing another example of a method for generating exclusion information. The input generation unit 120 in this example calculates the similarity between each piece of non-selected information and the selected information. The similarity between pieces of information may be calculated using a known method such as vectorization of text. The input generation unit 120 may calculate the similarity between pieces of information based on similarities in the words contained in each piece of information, the subject matter of the information, the issues or effects presented by the information, etc. The input generation unit 120 may calculate the similarity between each piece of non-selected information and each piece of selected information, and use the average value of the similarities for each piece of selected information.

[0059] The input generation unit 120 generates exclusion information corresponding to each piece of non-selected information based on the calculated similarity. For example, the input generation unit 120 may adjust the information granularity of the exclusion information generated from the non-selected information according to the similarity of the non-selected information. Information granularity is an index indicating the degree of fineness of the information. The higher the information granularity, the more detailed the information contains, and the lower the information granularity, the more abstract the information contains.

[0060] For non-selected information that has a low similarity to the selected information, it can be inferred that not only the details of the non-selected information but also the content related to the subject of the non-selected information is to be excluded from the summary information. On the other hand, for non-selected information that has a high similarity to the selected information, it can be inferred that the subject of the non-selected information may be included in the summary, but the details are not to be included.

[0061] For non-selected information whose similarity is higher than a reference value, the input generation unit 120 may extract words contained in the non-selected information as excluded information. For non-selected information whose similarity is lower than a reference value, the input generation unit 120 may extract the field to which the non-selected information belongs or the subject of the information as excluded information. For non-selected information whose similarity is low, the input generation unit 120 may generate excluded information that is a superordinate concept of words contained in the non-selected information.

[0062] In the example of FIG. 12, the input generation unit 120 generates, as excluded information, "information related to beauty," which is a superordinate concept of the words "anti-aging" and "beautiful skin" included in non-selected information that has a low similarity to the selected information. The excluded information indicates the subject of the non-selected information. On the other hand, for non-selected information that has a high similarity to the selected information, the input generation unit 120 extracts, as excluded information, one or more words included in the non-selected information. The input generation unit 120 may extract a larger number of words from the non-selected information as the similarity increases. According to this example, the excluded information is generated using the similarity between the non-selected information and the selected information, so that excluded information that more accurately reflects the user's intention can be generated.

[0063] The input generation unit 120 may calculate the similarity between non-selected information pieces and generate exclusion information based on the similarity. For example, for a group including multiple non-selected information pieces whose similarity to each other is higher than a reference value, exclusion information with lower information granularity (i.e., more abstract content) may be generated. For example, the input generation unit 120 may generate exclusion information for the group based on the common information theme, presentation task, presentation effect, etc., among the non-selected information pieces of the group. When multiple non-selected information pieces have high similarity, it can be inferred that the user wants to exclude the common theme, etc., among the non-selected information pieces from the summary information. Therefore, by using the similarity between non-selected information pieces, exclusion information that more accurately reflects the user's intentions can be generated.

[0064] 13 is a diagram showing another example of the configuration of the input generation device 100. The input generation device 100 of this example includes an information collection unit 170 that collects candidate information based on predetermined collection conditions. The input generation unit 120 of this example generates exclusion information based on the collection conditions. The configuration other than the input generation unit 120 and the information collection unit 170 is the same as that of the input generation device 100 of any of the aspects described in this specification.

[0065] The information collecting unit 170 collects multiple pieces of candidate information based on collection conditions input by the user. The collection conditions include keywords, phrases, sentences, etc. input by the user. The information collecting unit 170 collects candidate information that matches the collection conditions, for example, on the web. The information collecting unit 170 may collect candidate information from a specified database, etc. For example, the database may be a collection of the contents of patent specifications, academic papers, technical reports, newspaper articles, or any other specific type of information.

[0066] The presenting unit 102 presents to the user a plurality of pieces of candidate information collected by the information collecting unit 170. The classifying unit 110 classifies the plurality of pieces of candidate information into selected information and non-selected information. The operations of the presenting unit 102 and the classifying unit 110 are similar to those of the input generating device 100 according to any of the aspects described in this specification.

[0067] The input generating unit 120 generates exclusion information for each piece of non-selected information. The input generating unit 120 of this example generates the exclusion information based on the collection conditions in addition to the parameters used in any of the examples described in this specification.

[0068] For example, the input generation unit 120 may calculate the similarity between each piece of exclusion information generated without using the collection conditions and the collection conditions, and determine the exclusion information to be used in the input information based on the similarity. The similarity between the exclusion information and the collection conditions may be calculated using a known method such as vectorization of information. The input generation unit 120 may prioritize exclusion information with a lower similarity and use it as the input information. The input generation unit 120 may use exclusion information with a similarity equal to or less than a reference value as the input information. In another example, the input generation unit 120 may select exclusion information in ascending order of similarity so that the number of exclusion information to be included in the input information is equal to or less than a reference number. According to this example, information with a high similarity to the collection conditions is not included in the exclusion information. This makes it possible to generate exclusion information that accurately reflects the user's intention indicated by the collection conditions.

[0069] 14 is a flowchart illustrating an example of the operation of the input generation device 100. The input generation device 100 of this example extracts and lists exclusion information whose similarity is greater than a reference value. The input generation unit 120 may generate input information using the exclusion information in the list.

[0070] First, in collection condition acquisition step S402, the information collection unit 170 acquires collection conditions from the user. Next, in result file acquisition step S404, the information collection unit 170 acquires a result file including multiple pieces of candidate information collected according to the collection conditions.

[0071] Next, in the exclusion information acquisition stage S406, the input generation unit 120 acquires one or more pieces of exclusion information generated without using the collection conditions. The exclusion information may be the same as the exclusion information described in FIGS. 1 to 12.

[0072] The input generation unit 120 generates a list of exclusion information to be included in the input information by the processing in steps S408 to S428. In step S408, the input generation unit 120 sets a predetermined parameter i to 0. Next, in step S410, the input generation unit 120 determines whether the parameter i is smaller than the set value n. The set value n indicates the number of pieces of exclusion information acquired in the exclusion information acquisition stage S406. That is, step S410 determines whether the processing in S412 to S426 has been performed for each piece of exclusion information. If it is determined in step S410 that i < n is false (No), the processing for all pieces of exclusion information has been completed, so the processing after step S428 is performed. If it is determined in step S410 that i < n is true (Yes), there is unprocessed exclusion information remaining, so the processing after step S412 is performed.

[0073] In step S412, the input generation unit 120 acquires the vector representation Vec(ki) of the exclusion information ki acquired in S406. The exclusion information ki indicates the (i + 1)-th (where i is an integer from 0 to n - 1) piece of exclusion information among the exclusion information acquired in S406. The vector representation Vec(ki) is information obtained by vectorizing the content of the exclusion information ki. The vectorization of information can be executed by a known method using a language model or the like.

[0074] In step S414, the input generation unit 120 sets a predetermined parameter j to 0. Next, in step S416, the input generation unit 120 determines whether the parameter j is smaller than the set value m. The set value m indicates the number of collection conditions (for example, keywords). That is, step S416 determines whether the processing in S418 to S424 has been performed for each collection condition. If it is determined in step S416 that j < m is not true (No), since the processing for all collection conditions has ended, the processing after step S426 is performed. In step S426, the input generation unit 120 adds 1 to the parameter i and specifies the exclusion information ki to be processed next. If it is determined in step S416 that j < m is true (Yes), since unprocessed collection conditions remain, the processing after step S418 is performed.

[0075] In step S418, the input generation unit 120 acquires the vector representation Vec(qi) of the collection condition qi. The collection condition qi indicates the (j + 1)-th (where j is an integer from 0 to m - 1) collection condition among the collection conditions acquired in S402. The vector representation Vec(qi) is information obtained by vectorizing the content of the collection condition qi.

[0076] In step S420, the input generation unit 120 determines whether the similarity Sim(Vec(ki), Vec(qi)) between the vector representation Vec(ki) of the exclusion information ki and the vector representation Vec(qi) of the collection condition qi is greater than the reference value D. Here, as the similarity Sim, the cosine distance between Vec(ki) and Vec(qi) may be used. That is, if the similarity Sim is greater than the reference value D, it indicates that Vec(ki) and Vec(qi) are not similar. In this case, the processing after step S422 is performed. In step S422, the input generation unit 120 registers the exclusion information ki in the list KWL. Therefore, exclusion conditions that are not similar to the collection conditions are registered in the list KWL, and it is possible to prevent information of interest to the user from being excluded. If the similarity Sim is less than or equal to the reference value D, the input generation unit 120 performs the processing in step S424 without performing the processing in step S422.

[0077] In step S424, the input generation unit 120 adds 1 to the parameter j and specifies the collection condition qi to be processed next. The process from step S416 is repeated after step S424. That is, for the exclusion information ki to be processed, the similarity Sim with each collection condition qi is sequentially calculated and compared with the reference value D. In this example, when the similarity Sim with any of the collection conditions qi is greater than the reference value D, the exclusion information ki is registered in the list.

[0078] In step S416, if it is determined that j < m (that is, the processing of all collection conditions qi for the exclusion information ki has ended), the processing after step S426 is executed. In step S426, the input generation unit 120 adds 1 to the parameter i and specifies the exclusion information ki to be processed next.

[0079] After step S426, step S410 is executed. In step S410, the input generation unit 120 determines whether there is any unprocessed exclusion information ki remaining. If there is unprocessed exclusion information ki remaining, the input generation unit 120 repeats the processing after step S412. If the processing of all exclusion information ki has ended, the input generation unit 120 performs the processing of step S428.

[0080] In step S428, the input generation unit 120 outputs the list KWL. The input generation unit 120 may generate input information using the exclusion information ki registered in the list KWL. The input generation unit 120 may present the list KWL to the user. The input generation unit 120 may exclude the exclusion information specified by the user from the exclusion information included in the list KWL. The input generation unit 120 may generate input information based on the modified list KWL. <(

[0081] 15 is a diagram showing another example of the operation of the input generation device 100. The input generation unit 120 of this example acquires timing information relating to the creation time of each piece of non-selected information, and generates exclusion information corresponding to each piece of non-selected information based on the timing information. Operations other than the process of generating exclusion information using the timing information are the same as those of the input generation device 100 of any aspect described in this specification.

[0082] The time information indicates the time when the non-selected information was created. The time information may be the time when the non-selected information was published on the web, the time when the non-selected information was registered in a predetermined database, or the time when the creator of the non-selected information recorded it in association with the non-selected information. The time information may also indicate the last date when the non-selected information was updated.

[0083] The input generation unit 120 may generate exclusion information from each piece of non-selected information whose time information is older than a predetermined date. In another example, the input generation unit 120 may select non-selected information in order from the oldest time information so that the number of pieces of exclusion information to be extracted is equal to or less than a reference number, and generate exclusion information from the non-selected information. This allows the content of older non-selected information to be excluded from the summary information.

[0084] 16 is a flowchart illustrating an example of the operation of the input generation device 100. The input generation device 100 of this example creates a list of exclusion information generated from non-selected information with older timing information so that the number of pieces of exclusion information to be extracted is equal to or less than a reference number. The input generation unit 120 may generate input information using the exclusion information in the list.

[0085] The input generation unit 120 of this example sorts the non-selected information in chronological order according to the time information through the processes of steps S602 to S610. First, in step S602, the input generation unit 120 sets a predetermined parameter i to 0. Next, in step S604, the input generation unit 120 obtains a result file including one or more pieces of non-selected information and the time information of each piece of non-selected information.

[0086] Next, in step S606, the input generation unit 120 determines whether the parameter i is smaller than the set value n. The set value n in this example corresponds to the number of non-selected information included in the result file. That is, step S606 determines whether the processes in S608 to S610 have been performed for each non-selected information. If it is determined in step S606 that i < n is false (No), since the sorting process for all non-selected information has ended, the processes after step S612 are performed. If it is determined in step 606 that i < n is true (Yes), since there is still unsorted non-selected information, the processes after step S608 are performed.

[0087] In step S608, the input generation unit 120 adds 1 to the parameter i and specifies the non-selected information si to be sorted next. In step S610, the input generation unit 120 sorts the non-selected information si based on the time information. The input generation unit 120 may sort each non-selected information by comparing the time information between the sorted non-selected information and the non-selected information to be sorted.

[0088] When the sorting of all non-selected information is completed, the input generation unit 120 performs the processes after step S612. In step S612, the input generation unit 120 sets a predetermined parameter j to 0. Next, in step S614, the input generation unit 120 determines whether the parameter j is smaller than the set value m. The set value m indicates the number of exclusion information to be extracted. That is, in step S614, it is determined whether the number of exclusion information extracted in the processes after S616 has reached the set value m. If it is determined in step S616 that j < m is false (No), since the number of exclusion information has reached the set value m, the processes after step S620 are performed. If it is determined in step S614 that j < m is true (Yes), since the number of exclusion information has not reached the set value m, exclusion information is generated by the processes after step S616.

[0089] In step S616, the input generation unit 120 generates exclusion information from the j-th non-selected information sj. The method for generating exclusion information is the same as any of the examples described in this specification. In this example, exclusion information extracted from one piece of non-selected information is treated as one piece of exclusion information. In other words, even if multiple words are extracted from one piece of non-selected information, the group of words is treated as one piece of exclusion information. The input generation unit 120 registers the generated exclusion information in the list KWL.

[0090] In step S618, the input generation unit 120 adds 1 to the parameter j. That is, the input generation unit 120 specifies the next oldest non-selected information sj. The input generation unit 120 repeats the process from step S614. As described above, if it is determined in step S614 that the number of pieces of exclusion information registered in the list KWL has reached the set value m, the input generation unit 120 performs the process of step S620.

[0091] In step S620, the input generation unit 120 outputs the list KWL. The input generation unit 120 may generate input information using the exclusion information registered in the list KWL. The input generation unit 120 may present the list KWL to a user. The input generation unit 120 may exclude, from the list KWL, exclusion information specified by the user among the exclusion information included in the list KWL. The input generation unit 120 may generate input information based on the revised list KWL.

[0092] The summary generation system 10 described in Figures 1 to 16 may be realized by installing a program on one or more computers. The input generation device 100 described in Figures 1 to 16 may be realized by installing a program on one or more computers. The processing model 210 described in Figures 1 to 16 may be realized by installing a program on one or more computers. These programs may be recorded on computer-readable media.

[0093] Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where the blocks may represent (1) stages of a process in which operations are performed or (2) sections of an apparatus responsible for performing the operations. Particular stages and sections may be implemented by dedicated circuitry, programmable circuitry provided with computer-readable instructions stored on a computer-readable medium, and / or a processor provided with computer-readable instructions stored on a computer-readable medium. Dedicated circuitry may include digital and / or analog hardware circuitry, and may include integrated circuits (ICs) and / or discrete circuits. Programmable circuitry may include reconfigurable hardware circuitry, including logical AND, OR, XOR, NAND, NOR, and other logical operations, flip-flops, registers, memory elements such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like.

[0094] A computer-readable medium may include any tangible device capable of storing instructions that are executed by an appropriate device, such that the computer-readable medium having instructions stored thereon comprises an article of manufacture containing instructions that can be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of computer-readable media may include electronic, magnetic, optical, electromagnetic, and semiconductor storage media. More specific examples of computer-readable media may include floppy disks, diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disc read-only memory (CD-ROM), digital versatile disc (DVD), Blu-ray (RTM) disc, memory stick, integrated circuit card, and the like.

[0095] The computer readable instructions may include either assembler instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages ​​such as the “C” programming language or similar programming languages.

[0096] The computer-readable instructions may be provided to a processor or programmable circuit of a programmable data processing device, such as a general-purpose computer, a special-purpose computer, or another computer, either locally or via a wide-area network (WAN) such as a local area network (LAN) or the Internet, which executes the computer-readable instructions to create means for performing the operations specified in the flowcharts or block diagrams. Here, the computer may be a personal computer (PC), a tablet computer, a smartphone, a workstation, a server computer, a general-purpose computer, a special-purpose computer, or the like, or may be a computer system in which multiple computers are connected. Such a computer system in which multiple computers are connected is also called a distributed computing system, and is a broad definition of computers. In a distributed computing system, multiple computers collectively execute a program by each executing a portion of the program and passing data between computers as needed during program execution.

[0097] Examples of processors include a computer processor, a central processing unit (CPU), a processing unit, a microprocessor, a digital signal processor, a controller, a microcontroller, etc. A computer may have one or more processors. In a multiprocessor system with multiple processors, each processor executes a portion of a program and passes data between processors as needed during program execution, allowing the multiple processors to collectively execute a program. For example, in multitasking, each of the multiple processors may execute a portion of each task in small chunks by switching tasks at time slice intervals. In this case, which portion of a program each processor executes changes dynamically. Which portion of a program each of the multiple processors executes may also be statically determined by multiprocessor-aware programming.

[0098] 17 illustrates an example of a computer 1200 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 1200 may cause the computer 1200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and / or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 1212 to cause the computer 1200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0099] A computer 1200 according to this embodiment includes a CPU 1212, a RAM 1214, a graphics controller 1216, and a display device 1218, which are interconnected by a host controller 1210. The computer 1200 also includes input / output units such as a communication interface 1222, a storage device 1224 such as a hard disk drive, a DVD-ROM drive 1226, and an IC card drive, which are connected to the host controller 1210 via an input / output controller 1220. The computer also includes legacy input / output units such as a ROM 1230 and a keyboard 1242, which are connected to the input / output controller 1220 via an input / output chip 1240.

[0100] The CPU 1212 operates according to programs stored in the ROM 1230 and the RAM 1214, thereby controlling each unit. The graphics controller 1216 acquires image data generated by the CPU 1212 into a frame buffer or the like provided in the RAM 1214 or into the graphics controller 1216 itself, and causes the image data to be displayed on the display device 1218.

[0101] The communication interface 1222 communicates with other electronic devices via a network. The storage device 1224 stores programs and data used by the CPU 1212 in the computer 1200. The DVD-ROM drive 1226 reads programs or data from a DVD-ROM 1227 and provides the programs or data to the storage device 1224 via the RAM 1214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0102] The ROM 1230 stores therein a boot program or the like that is executed by the computer 1200 upon activation, and / or programs that depend on the hardware of the computer 1200. The input / output chip 1240 may also connect various input / output units to the input / output controller 1220 via a parallel port, a serial port, a keyboard port, a mouse port, etc.

[0103] The programs are provided by a computer-readable medium such as a DVD-ROM 1227 or an IC card. The programs are read from the computer-readable medium, installed in the storage device 1224, RAM 1214, or ROM 1230, which are also examples of computer-readable media, and executed by the CPU 1212. Information processing described in these programs is read by the computer 1200, and causes cooperation between the programs and the various types of hardware resources described above. An apparatus or a method may be configured by implementing information manipulation or processing in accordance with the use of the computer 1200.

[0104] For example, when communication is performed between the computer 1200 and an external device, the CPU 1212 may execute a communication program loaded into the RAM 1214 and instruct the communication interface 1222 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 1212, the communication interface 1222 reads transmission data stored in a transmission buffer processing area provided in the RAM 1214, the storage device 1224, the DVD-ROM 1227, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes reception data received from the network to a reception buffer processing area or the like provided on the recording medium.

[0105] The CPU 1212 may read all or a necessary portion of a file or database stored on an external recording medium such as the storage device 1224, the DVD-ROM drive 1226 (DVD-ROM 1227), an IC card, etc. into the RAM 1214, and perform various types of processing on the data on the RAM 1214. The CPU 1212 then writes back the processed data to the external recording medium.

[0106] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. CPU 1212 may perform various types of processing on data read from RAM 1214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequences of the programs, and write the results back to RAM 1214. CPU 1212 may also search for information in a file, database, etc. on the recording medium. For example, if multiple entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored on the recording medium, CPU 1212 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0107] The above-described programs or software modules may be stored in a computer-readable medium on or near the computer 1200. A recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can also be used as a computer-readable medium, thereby providing the programs to the computer 1200 via the network.

[0108] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

[0109] It should be noted that the execution order of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and that the processes can be performed in any order unless the output of a previous process is used in a subsequent process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that the processes must be performed in this order. [Explanation of symbols]

[0110] 10. Summary generation system, 100. Input generation device, 102. Presentation unit, 104. Input part, 110. Classification unit, 120. Input generation unit, 130. Summary acquisition unit, 150. Exclusion information editing unit, 160. History storage unit, 170. Information collection unit, 200. Summary generation unit, 210. Processing model

Claims

1. An input generation device that generates input information to be input to a processing model, in order to cause the processing model to generate summary information of a plurality of candidate information, a classification unit that classifies the plurality of pieces of candidate information into one or more pieces of selected information selected by a user and one or more pieces of non-selected information not selected by the user; an input generation unit that generates the input information based on the content of one or more pieces of selection information and the content of one or more pieces of non-selection information; An input generating device comprising:

2. The input generating unit generates one or more pieces of exclusion information indicating items included in the non-selected information, and generates the input information for generating the summary information by excluding the exclusion information from the content of the selected information. The input generating device of claim 1 .

3. The input generation unit extracts at least a part of words included in the non-selected information as the excluded information. The input generating device of claim 2 .

4. an information collection unit that collects the plurality of pieces of candidate information based on collection conditions input by the user; a presentation unit that presents the collected plurality of pieces of candidate information to the user; Further provided with The input generation unit generates the exclusion information based on the collection conditions. The input generating device of claim 2 .

5. The input generation unit presents the exclusion information generated from the non-selection information to the user, acquires an instruction from the user regarding the exclusion information, and determines the exclusion information to be included in the input information. The input generating device of claim 2 .

6. further comprising a history storage unit that records the user's past instructions; The input generation unit generates the exclusion information based on the past instructions of the same user. The input generating device of claim 5 .

7. The input generation unit calculates a similarity between each of the pieces of exclusion information and the selected information, and presents the user with one or more pieces of exclusion information arranged based on the calculated similarity. The input generating device of claim 5 .

8. The input generation unit calculates a similarity between each piece of exclusion information and the selected information, and determines the exclusion information to be included in the input information based on the calculated similarity. An input generating device according to any one of claims 2 to 6.

9. The input generation unit calculates a similarity between each piece of non-selected information and the selected information, and generates the exclusion information corresponding to each piece of non-selected information based on the calculated similarity. An input generating device according to any one of claims 2 to 6.

10. The input generation unit acquires time information relating to a creation time of each of the non-selected information, and generates the exclusion information corresponding to each of the non-selected information based on the time information. An input generating device according to any one of claims 2 to 6.

11. The input generating unit presents the exclusion information and one or more provisional summaries generated without using the exclusion information based on the selection information to the user, and generates the input information based on an instruction obtained from the user regarding the exclusion information. An input generating device according to any one of claims 2 to 6.

12. A summary generation system for generating summary information of a plurality of candidate information, a classification unit that classifies the plurality of pieces of candidate information into one or more pieces of selected information selected by a user and one or more pieces of non-selected information not selected by the user; a summary generation unit that generates the summary information based on the content of one or more pieces of selected information and the content of one or more pieces of non-selected information; A summary generation system comprising:

13. A program for causing a computer to function as the input generating device according to claim 1.

Images