Reading comprehension support methods

Abstract

We provide a document reading support system that accurately presents the information necessary for the user. [Solution] A reading comprehension support system is provided that receives a specified document, creates a first graph representing the structure of the specified document using the words contained in the specified document, outputs two or more words contained in the first graph, receives multiple specified words from the outputted words, searches the first graph using the multiple specified words, and outputs the search results. As a search result, at least a second graph showing the shortest path between any two of the multiple specified words in the first graph can be output. The shortest path is a path that connects any two of the multiple specified words via at least one complementary word. The complementary word is a word different from the multiple specified words.

Description

Technical Field

[0001] One aspect of the present invention relates to a document reading support system and a reading support method.

[0002] Note that one aspect of the present invention is not limited to the above technical field. Examples of the technical field of one aspect of the present invention include semiconductor devices, display devices, light-emitting devices, power storage devices, storage devices, electronic devices, lighting devices, input devices (e.g., touch sensors, etc.), input / output devices (e.g., touch panels, etc.), their driving methods, or their manufacturing methods.

Background Art

[0003] When reading a document, the way of reading the document varies depending on the purpose of the reader or the type of the document. Sometimes the entire document is read, and sometimes the purpose is to search for information necessary for the reader, and it may be sufficient to search for the location where the necessary information is described in the document and only glance at the relevant location. As a method of searching for necessary information in a document, there is a method of using a table of contents or an index. For an electronic document, there is also a method of searching for desired information by searching for a keyword. In addition, a method of performing structural analysis of a document according to set rules has been proposed (Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] When using a table of contents or index, it is inefficient if the word you want to find is not used in the table of contents or index. Text search using keywords allows you to find sentences or paragraphs containing keywords throughout the entire document, but it is not always efficient to find the desired information. Reasons for this inefficiency include too many matching locations for the keyword, making it time-consuming to reach the desired information, not being able to narrow down the desired information with a single keyword, or not being able to find a suitable keyword. Furthermore, when performing structural analysis of a document according to rules, the structure of the document being read is limited, making it difficult to handle documents with various structures. One aspect of the present invention solves at least one of these problems.

[0006] One aspect of the present invention aims to provide a document reading support system or method that accurately presents necessary information to the user. Another aspect of the present invention aims to provide a document reading support system or method that assists the user in understanding a document. Another aspect of the present invention aims to provide a document reading support system or method that is easy for the user to operate.

[0007] Furthermore, the description of these problems does not preclude the existence of other problems. One aspect of the present invention does not necessarily have to solve all of these problems. It is possible to extract other problems from the description in the specification, drawings, and claims. [Means for solving the problem]

[0008] One aspect of the present invention is a reading comprehension support system having a reception unit, a processing unit, and an output unit. The reception unit has a function to receive a specified document and a function to receive a plurality of specified words. The processing unit has a function to create a first graph representing the structure of a specified document using words contained in the specified document and a function to search the first graph using a plurality of specified words. The output unit has a function to output a plurality of words contained in the first graph and a function to output the search results of the first graph. The plurality of specified words are at least a portion of the plurality of words contained in the first graph.

[0009] Preferably, the output unit outputs a second graph as a search result, which shows at least the shortest path between any two of the multiple specified terms in the first graph. Preferably, the output unit has a function to output sentences containing the specified terms in paragraphs in the specified document that contain two or more of the multiple specified terms. The shortest path is a path that connects any two of the multiple specified terms via at least one complementary term, and preferably the complementary term is a term different from the multiple specified terms. Preferably, the output unit has a function to output sentences containing at least one of the specified term and the complementary term in paragraphs in the specified document that contain at least one of the multiple specified terms and at least one complementary term.

[0010] Alternatively, the output unit preferably outputs, as a search result, a second graph showing at least the shortest path between each of the multiple specified terms in the first graph. The output unit preferably has a function to output sentences containing the specified terms in paragraphs in the specified document that contain two or more of the multiple specified terms. The shortest path connecting any two of the multiple specified terms is a path that connects the two specified terms via at least one complementary term, and the complementary term is preferably a term different from the multiple specified terms. The output unit preferably has a function to output sentences containing at least one of the specified term and the complementary term in paragraphs in the specified document that contain at least one of the multiple specified terms and at least one complementary term.

[0011] A reading comprehension support system according to one aspect of the present invention preferably further includes a storage unit for storing search results.

[0012] One aspect of the present invention is a reading comprehension support method that receives a designated document, creates a first graph representing the structure of the designated document using words and phrases contained in the designated document, outputs two or more words and phrases contained in the first graph, receives multiple designated words and phrases from the outputted words and phrases, searches the first graph using the multiple designated words and phrases, and outputs the search results.

[0013] Preferably, as a search result, a second graph is output that shows the shortest path between any two of the multiple specified terms in the first graph. Preferably, along with the search result, sentences containing the specified terms in paragraphs of the specified document that contain two or more of the multiple specified terms are output. The shortest path is a path that connects any two of the multiple specified terms via at least one complementary term, and preferably the complementary term is a term different from the multiple specified terms. Preferably, along with the search result, sentences containing at least one of the specified term and the complementary term in paragraphs of the specified document that contain at least one of the multiple specified terms and at least one complementary term are output.

[0014] Alternatively, it is preferable to output, as a search result, a second graph showing the shortest path between each of the multiple specified terms in the first graph. It is preferable to output, along with the search results, sentences containing the specified terms in paragraphs in the specified document that contain two or more of the multiple specified terms. The shortest path connecting any two of the multiple specified terms is a path connecting the two specified terms via at least one complementary term, and it is preferable that the complementary term is a different term from the multiple specified terms. It is preferable to output, along with the search results, sentences containing at least one of the specified term and the complementary term in paragraphs in the specified document that contain at least one of the multiple specified terms and at least one complementary term. [Effects of the Invention]

[0015] According to one aspect of the present invention, it is possible to provide a document reading support system or a document reading support method that accurately presents information necessary for a user. According to one aspect of the present invention, it is possible to provide a reading support system or a reading support method that supports a user in understanding a document. According to one aspect of the present invention, it is possible to provide a document reading support system or a document reading support method that is easy for a user to operate.

[0016] Note that the description of these effects does not prevent the existence of other effects. One aspect of the present invention does not necessarily have to have all of these effects. It is possible to extract other effects from the descriptions in the specification, drawings, and claims.

Brief Description of the Drawings

[0017] [Figure 1] FIG. 1 is a diagram showing an example of a reading support system. [Figure 2] FIG. 2 is a diagram showing an example of a reading support method. [Figure 3] FIGS. 3A to 3D are diagrams showing an example of a reading support method. [Figure 4] FIGS. 4A to 4E are diagrams showing an example of a reading support method. [Figure 5] FIGS. 5A to 5C are diagrams showing an example of a graph. [Figure 6] FIG. ⑥ is a diagram showing an example of output content. [Figure 7] FIG. 7 is a diagram showing an example of a graph. [Figure 8] FIG. 8 is a diagram showing an example of a reading support system. [Figure 9] FIG. 9 is a diagram showing an example of a reading support system.

Mode for Carrying Out the Invention

[0018] The embodiments will be described in detail with reference to the drawings. However, the present invention is not limited to the following description, and it will be easily understood by those skilled in the art that the form and details thereof can be variously changed without departing from the spirit and scope of the present invention. Therefore, the present invention is not to be construed as being limited to the description of the embodiments shown below.

[0019] In the configuration of the invention described below, the same reference numerals are commonly used among different drawings for the same part or parts having the same or similar functions, and the repeated description thereof will be omitted. Also, when referring to similar functions, the hatch patterns may be the same, and there may be cases where no reference numerals are particularly assigned.

[0020] Also, the positions, sizes, ranges, etc. of each configuration shown in the drawings may not represent the actual positions, sizes, ranges, etc. for the sake of easy understanding. Therefore, the disclosed invention is not necessarily limited to the positions, sizes, ranges, etc. disclosed in the drawings.

[0021] Note that the terms "film" and "layer" can be interchanged with each other depending on the case or situation. For example, the term "conductive layer" can be changed to the term "conductive film". Or, for example, the term "insulating film" can be changed to the term "insulating layer".

[0022] (Embodiment 1) In this embodiment, a reading support system and a reading support method according to an aspect of the present invention will be described with reference to FIGS. 1 to 7.

[0023] In a reading support system according to an aspect of the present invention, a designated document is received, a first graph representing the structure of the designated document is created using the words and phrases included in the designated document, and two or more words and phrases included in the first graph are output. Then, a plurality of designated words and phrases are received from the output words and phrases, the first graph is searched using the plurality of designated words and phrases, and the search result is output. Note that, in this specification and the like, a graph can also be referred to as a graph structure.

[0024] In creating the first graph, words and phrases that are located close together in a document can be directly connected. For example, if two words and phrases exist in the same sentence, they can be directly connected. Also, if two words and phrases exist in the same paragraph, they can be directly connected. Furthermore, if, for two words and phrases, the sentence containing one word and phrase is close to the sentence containing the other word and phrase (for example, within n sentences before or after, where n is an integer greater than or equal to 1), the two words and phrases can be directly connected. In this way, by connecting words and phrases that are located close together in a document, a graph showing the structure of the document can be created. By creating such a graph, the relationships between each word and phrase in the document can be shown.

[0025] The user of the reading comprehension support system specifies the document they want to read as the "specified document." The user also specifies multiple keywords related to the information they want to obtain as "specified vocabulary."

[0026] When simply performing a keyword search on a document, the reader is required to select keywords to use for the search, taking into account synonyms, related terms, and variations in spelling. Therefore, keyword selection can be burdensome for the reader, and differences in skill are likely to occur. On the other hand, one embodiment of the present invention, a reading support system, receives a specified document, creates a first graph, and then outputs the words and phrases contained in the first graph. The user of the reading support system can select keywords from the outputted words and phrases. Therefore, keyword selection is easy, differences in user skill are less likely to occur, and necessary information can be quickly found from the document.

[0027] Furthermore, even if a reader selects multiple keywords, these keywords may be scattered throughout the document, making it difficult to understand the relationships between them. For example, even if a reader uses a book's index to find information on several keywords, the content may not connect. This can lead to increased search and comprehension time, requiring the reader to add more keywords or read between pages.

[0028] One embodiment of the present invention is a reading comprehension support system that can output a second graph showing the relationships between multiple specified terms by searching a first graph using multiple specified terms received. This allows the user to easily grasp the relationships between the specified terms. Furthermore, one embodiment of the present invention is a reading comprehension support system that can extract and output sentences containing multiple specified terms specified by the user. By reading the extracted sentences, the user can efficiently obtain the necessary information.

[0029] A reading comprehension support system according to one aspect of the present invention can present the shortest path between each of several specified words in a first graph. For example, by outputting a second graph showing the shortest path, the system can present the relationships between several specified words to the user.

[0030] For example, the shortest path between the first and second specified terms may include other specified terms. Users can understand the relationships between multiple specified terms and deepen their understanding of the document.

[0031] Furthermore, the shortest path may include supplementary terms that are different from the specified terms. By presenting supplementary terms that the user has not specified, it is possible to facilitate the understanding and comprehension of the document's content. Users can deepen their understanding of the document by understanding the supplementary terms themselves, and furthermore, by understanding the relationship between the supplementary terms and the specified terms. Supplementary terms are terms included in the specified document (i.e., terms included in the first graph), and are different from the specified terms.

[0032] A reading comprehension support system according to one aspect of the present invention can output sentences containing specified words or phrases from a specified document, along with a second graph. In this case, for example, all sentences containing any of the specified words or phrases can be output. However, depending on the specified words or phrases, the output may be too numerous, and it may take time for the user to find the information they want.

[0033] Therefore, in one aspect of the present invention, the reading comprehension support system preferably extracts and outputs sentences from a document based on each shortest path.

[0034] For example, it is possible to output sentences containing specified terms in paragraphs that contain two or more of the specified terms in a given document. Also, for example, it is possible to output sentences containing at least one of the specified terms and at least one of the complementary terms in paragraphs that contain at least one of the specified terms and at least one of the complementary terms in a given document.

[0035] This allows users to efficiently identify the sentences necessary to understand the relationships between multiple specified terms, and to quickly obtain the information they need.

[0036] Furthermore, one embodiment of the present invention provides the shortest path between at least two of a plurality of designated words. In other words, one embodiment of the present invention may provide the shortest path between some of the designated words, or it may provide the shortest path between all of the designated words.

[0037] For example, there may be cases where two specified terms cannot be connected even through other terms, making it impossible to show a path. Alternatively, for example, a criterion for determining the degree of relevance between two specified terms may be established, and if the system determines that the two specified terms are highly related, it may present the shortest path between them. Specifically, if the shortest path between two specified terms is connected through a predetermined number of terms or fewer, it can be determined that the two specified terms are not highly related. Conversely, if the shortest path between two specified terms is connected through more than a predetermined number of terms, it can be determined that the two specified terms are not highly related.

[0038] A reading comprehension support system according to one aspect of the present invention can also be used for proofreading documents. For example, among the specified terms, there may be isolated terms that do not connect with other specified terms. In this case, the reading comprehension support system according to one aspect of the present invention may output the terms that do not connect with other specified terms as isolated terms. Also, the content of the output graph may differ from what was expected, such as when related specified terms are not connected. In this case, there may be errors or omissions in the document. Thus, by using the reading comprehension support system according to one aspect of the present invention, documents can be reviewed efficiently.

[0039] Furthermore, a reading comprehension support system according to one aspect of the present invention can also be used to grasp either or both the relationships and differences between multiple documents. For example, a reading comprehension support system according to one aspect of the present invention can create a first graph representing the structure of each of several designated documents using the words contained in each designated document, search each of the first graphs, and output the search results. By comparing the output results, the user can easily confirm the relationships and differences between the multiple documents.

[0040] Furthermore, a reading comprehension support system according to one aspect of the present invention may have a function to compare search results for multiple documents and present at least one of the relationships and differences. For example, a reading comprehension support system according to one aspect of the present invention can create a graph showing the shortest path between specified words in each document as a search result. Then, by vectorizing the graph and calculating the similarity of each vector, the similarity of multiple documents can be evaluated.

[0041] In this case, two or more terms included in each first graph may be output, and specified terms may be accepted for each specified document. Alternatively, specified terms common to all specified documents may be accepted. Furthermore, if a synonym or equivalent term exists for a term included in one specified document in another specified document, it is preferable to link these terms. For example, if "insulating film" and "insulating layer" are linked, and "insulating film" is selected as the specified term, then in one specified document, the graph may be searched using "insulating film," and in another specified document, the graph may be searched using "insulating layer."

[0042] <Reading Comprehension Support System 1> Figure 1 shows a block diagram of the reading comprehension support system 100. The reading comprehension support system 100 includes a reception unit 110, a storage unit 120, a processing unit 130, an output unit 140, and a transmission line 150.

[0043] The reading comprehension support system 100 may be installed on an information processing device such as a personal computer used by the user. Alternatively, the processing unit of the reading comprehension support system 100 may be installed on a server, and it may be used by accessing it from a client PC via a network.

[0044] [Reception Desk 110] The reception unit 110 receives specified documents. The reception unit also receives specified phrases. The data supplied to the reception unit 110 is supplied via the transmission line 150 to one or both of the storage unit 120 and the processing unit 130.

[0045] Unless otherwise specified in this specification, a document is a description of an event in natural language that is digitized and machine-readable. Documents include, but are not limited to, patent applications, case law, contracts, terms and conditions, product manuals, novels, publications, white papers, and technical documents.

[0046] [Storage section 120] The storage unit 120 has the function of storing the program executed by the processing unit 130. Preferably, the storage unit 120 also has the function of storing the graph generated by the processing unit 130. It is desirable that the graph be linked to a document so that it is clear which document it was created from. The storage unit 120 may also have the function of storing the calculation results and inference results generated by the processing unit 130, as well as the data input to the reception unit 110.

[0047] The storage unit 120 includes at least one of volatile memory and non-volatile memory. Examples of volatile memory include DRAM (Dynamic Random Access Memory) and SRAM (Static Random Access Memory). Examples of non-volatile memory include ReRAM (Resistive Random Access Memory), PRAM (Phase-change Random Access Memory), FeRAM (Ferroelectric Random Access Memory), MRAM (Magnetoresistive Random Access Memory), and flash memory. The storage unit 120 may also include a recording media drive. Examples of recording media drives include hard disk drives (HDDs) and solid state drives (SSDs).

[0048] The storage unit 120 may have a database containing document data.

[0049] Furthermore, the reading comprehension support system 100 may have a function to retrieve document data from a database located outside the system. For example, the reading comprehension support system may have a function to retrieve data from a database located outside the system.

[0050] Furthermore, the reading comprehension support system 100 may have the function to retrieve data from both its own database and an external database.

[0051] The database can be configured to include, for example, text data and / or image data.

[0052] Alternatively, storage and / or a file server may be used instead of a database. For example, if files on a file server are used, it is preferable that the database contains the paths to the files stored on the file server.

[0053] For example, a database could be an application database. Examples of applications include intellectual property applications such as patent applications, utility model registration applications, and design registration applications. There are no limitations on the status of each application; whether it has been published, is pending at the Japan Patent Office, or is registered is irrelevant. For instance, an application database could contain at least one of the following: applications awaiting examination, applications under examination, and registered applications, and could contain all of them.

[0054] For example, the application database preferably contains one or both of the specifications and claims for multiple patent applications. The specifications and claims are stored, for example, as text data.

[0055] The application database may contain at least one of the following: an application management number (including a proprietary number within the company) to identify an application, an application family management number to identify an application family, an application number, a publication number, a registration number, drawings, an abstract, a filing date, a priority date, a publication date, a status, a classification (such as a patent classification or utility model classification), a category, and keywords. Each of these pieces of information may be used to identify a document when a designated document is received. Alternatively, each of these pieces of information may be output along with the processing result of the processing unit 130.

[0056] In addition, various types of documents, such as books, magazines, newspapers, and academic papers, can be managed in the database. The database contains at least the text data of the documents. The database may also contain at least one of the following to identify each document: a number, title, publication date, author, and publisher. Each of these pieces of information may be used to identify a document when a specified document is received. Alternatively, each of these pieces of information may be output along with the processing result of the processing unit 130.

[0057] [Processing step 130] The processing unit 130 has the function of performing calculations and inferences using data supplied from either or both of the receiving unit 110 and the storage unit 120. The processing unit 130 also has the function of performing processing using various data contained in the database. The processing unit 130 can supply processing results, such as calculation results and inference results, to either or both of the storage unit 120 and the output unit 140.

[0058] The processing unit 130 has the function of performing morphological analysis. In other words, the processing unit 130 has the function of dividing each sentence contained in a document into the smallest units that have meaning in language (also called tokens, morphemes, words, etc.) and determining the part of speech of each token. The process of dividing each sentence into the smallest units can also be called lexical analysis.

[0059] The processing unit 130 preferably has a function for performing compound word analysis. In other words, it is preferable that it has a function for performing morphological analysis while considering compound words (such as compound nouns). For example, it is preferable that the processing unit 130 has a function for generating a new token whose part of speech is a compound noun (redefining a token) by combining several tokens in order to group consecutive nouns in a sentence together. Note that even if the part of speech of a token is a compound noun, the part of speech of that token may simply be described as a noun.

[0060] Furthermore, it is preferable that the processing unit 130 has a function to calculate the distance between each token. For example, it is preferable that the processing unit 130 can obtain information such as whether two tokens are in the same sentence or the same paragraph. It is also preferable that the processing unit 130 can calculate how many paragraphs, sentences, words, or strings of characters separate the two tokens.

[0061] Furthermore, it is preferable that the processing unit 130 has a function to obtain related words for each token. Related words include synonyms, equivalent words, superordinate words, and subordinate words. It is also preferable that the processing unit 130 has a function to calculate the similarity between each token.

[0062] Related words can be obtained, for example, from a dictionary such as a conceptual dictionary. This dictionary may be included in the reading comprehension support system or it may be provided externally. A conceptual dictionary is a list that includes word classifications and relationships with other words. The conceptual dictionary may be an existing conceptual dictionary, or a conceptual dictionary specific to the document's field may be created, or words that are likely to be used in the document's field may be added to a general-purpose conceptual dictionary.

[0063] Alternatively, the words may be vectorized (converted to numerical values), and one or both of the similarity and distance between multiple words may be calculated. Based on the degree of similarity or proximity between multiple words, related words for the node may be obtained.

[0064] Methods for determining the similarity between two vectors include cosine similarity, covariance, unbiased covariance, and Pearson's product-moment correlation coefficient. Of these, cosine similarity is particularly preferred.

[0065] Methods for calculating the distance between two vectors include the Euclidean distance, standardized (mean) Euclidean distance, Mahalanobis distance, Manhattan distance, Chebyshev distance, and Minkowski distance.

[0066] For example, it is preferable to generate distributed representation vectors of words using machine learning. It is even more preferable to generate distributed representation vectors of words using a neural network. Specifically, it is preferable to extract related words using distributed representation vectors obtained by machine learning the distributed representations of words contained in a specified document. Alternatively, it is preferable to extract related words using distributed representation vectors obtained by machine learning the distributed representations of words contained in a group of documents contained in a database or the like.

[0067] Furthermore, the processing unit 130 may have a function to calculate the frequency of occurrence of each token. For example, it is preferable to calculate the TF (Term Frequency) value of each token. The TF value can represent the frequency of occurrence of each token within a specified document.

[0068] Furthermore, the processing unit 130 may have a function to calculate the importance of each token. For example, it is preferable to calculate the TF-IDF (Term Frequency-Inverse Document Frequency) value of each token. The IDF value represents the degree to which a token appears concentrated in some documents. Tokens that appear in many documents have a small IDF value, while tokens that appear in only some documents have a large IDF value. For example, it is preferable to calculate the IDF value of a token using documents included in a database. By calculating the product of the TF value and the IDF value of each token, a score can be calculated to determine whether the token is a token that characterizes a specified document.

[0069] The processing unit 130 has the function of creating a graph that represents the structure of a document using the words and phrases contained in the document.

[0070] A graph has nodes (vertices) and edges. Nodes and edges can each have labels. The tokens mentioned above can be used as labels for nodes. For example, a token whose part of speech is a noun (including compound nouns) can be used as a node label. For edge labels, the distance between each of the tokens mentioned above, and one or both of the related words of each token, can be used.

[0071] As a graph, you may create either a directed graph using edges with direction, or an undirected graph using edges without direction.

[0072] Multiple nodes are connected by edges. The edges between two nodes may be single or multiple. When displaying a graph, edges can be represented using either straight lines, curves, or both.

[0073] Furthermore, the structure of a single document may be represented by multiple graphs. For example, the structure of a single document may be represented using both directed and undirected graphs.

[0074] Edges without direction are preferable to connect two nodes so that the relationship between the two nodes in the document can be understood. Conditions for connecting nodes include connecting nodes in the same sentence with an edge, connecting nodes in the same paragraph with an edge, or connecting nodes within a predetermined distance (e.g., a certain number of words or characters) with an edge.

[0075] When creating a directed graph, it is preferable that the processing unit 130 has the function of performing syntactic analysis. In other words, it is preferable that the processing unit 130 has the function of dividing each sentence contained in the document into tokens, determining the part of speech of each token, and determining the dependency relationships of each token. Some of the processing included in syntactic analysis can also be called lexical analysis or morphological analysis as described above. By performing syntactic analysis, the direction of dependency relationships can be indicated by arrows in the directed graph.

[0076] In creating a directed graph, for example, edges may be directed from earlier-appearing nodes to later-appearing nodes. Alternatively, the direction of edges may be determined based on dependency relationships obtained through syntactic analysis, hyperhypo-hypohypohypohypohypo, frequency of occurrence, or word importance.

[0077] The graph may be created based on rules derived from the dependency relationships between tokens. Alternatively, the graph may be created using a pre-trained model based on machine learning. For example, a conditional random field (CRF) may be used to perform machine learning that labels nodes and edges based on a list of tokens. This allows for labeling nodes and edges based on a list of tokens. Alternatively, a recurrent neural network (RNN), long short-term memory (LSTM), etc., may be used to train a Seq2Seq model that takes a list of tokens as input and outputs the orientation of nodes and edges. This allows for outputting the orientation of nodes and edges from a list of tokens.

[0078] Furthermore, the processing unit 130 has a function to search the created graph. For example, the processing unit 130 can find the shortest path between each of multiple words. Methods for finding the shortest path include Dijkstra's algorithm, Bellman-Ford algorithm, and Warshall-Floyd algorithm. For example, the path with the fewest number of included nodes (words) can be considered the shortest path.

[0079] Furthermore, the processing unit 130 has the function of creating a graph that shows the shortest path between each of several specified terms. The graph created by the processing unit 130 is output by the output unit 140.

[0080] Furthermore, it is preferable that the processing unit 130 has a function to vectorize the graph resulting from the search (for example, a graph showing the shortest path between each of several specified terms). Examples of methods for vectorizing graphs include the Weisfeiler-Lehman kernel.

[0081] Furthermore, it is preferable that the processing unit 130 has a function to calculate the similarity of vectors. This allows the graph, which is the search result of multiple documents, to be vectorized and the similarity of the multiple documents to be calculated.

[0082] Furthermore, when determining the similarity between multiple documents, it is sometimes possible to obtain a highly accurate similarity score by using a graph created through token abstraction. By abstracting tokens, documents can be grasped conceptually. Therefore, the similarity score can be calculated based on the concept of the document, and is less affected by the structure and expression of the document.

[0083] On the other hand, in order for users to accurately understand a document, it is preferable to present the words and phrases used in the document themselves. Therefore, the processing unit 130 may create both a graph created without abstracting tokens for comprehension support and a graph created with abstracted tokens for similarity calculation.

[0084] Token abstraction refers to replacing tokens with representative terms or hypernomies. Representative terms and hypernomies can be obtained using a conceptual dictionary or by machine learning classification. Token abstraction can be performed, for example, by vectorizing tokens using the morphemes they contain and classifying them using a classifier. Algorithms such as decision trees, support vector machines, random forests, and multilayer perceptrons may be used as the classifier. Specifically, it is good practice to classify "oxide semiconductors," "amorphous semiconductors," "silicon semiconductors," and "GaAs semiconductors" as "semiconductors." Additionally, it is good practice to classify "oxide semiconductor layers," "oxide semiconductor films," "amorphous semiconductor layers," "amorphous semiconductor films," "silicon semiconductor layers," "silicon semiconductor films," "GaAs semiconductor layers," and "GaAs semiconductor films" as "semiconductors."

[0085] The processing unit 130 may, for example, have an arithmetic circuit. The processing unit 130 may, for example, have a central processing unit (CPU).

[0086] The processing unit 130 may have a microprocessor such as a DSP (Digital Signal Processor) or a GPU (Graphics Processing Unit). The microprocessor may be implemented using a PLD (Programmable Logic Device) such as an FPGA (Field Programmable Gate Array) or FPAA (Field Programmable Analog Array). The processing unit 130 can perform various data processing and program control by interpreting and executing instructions from various programs via the processor. Programs that can be executed by the processor are stored in at least one of the processor's memory area and the storage unit 120.

[0087] The processing unit 130 may have main memory. The main memory includes at least one of volatile memory such as RAM (Random Access Memory) and non-volatile memory such as ROM (Read Only Memory).

[0088] For RAM, for example, DRAM or SRAM is used, and a virtual memory space is allocated and used as the workspace for the processing unit 130. The operating system, application programs, program modules, program data, and lookup tables stored in the storage unit 120 are loaded into RAM for execution. These data, programs, and program modules loaded into RAM are directly accessed and manipulated by the processing unit 130.

[0089] ROM can store BIOS (Basic Input / Output System) and firmware, etc., which do not require rewriting. Examples of ROM include mask ROM, OTPROM (One Time Programmable Read Only Memory), and EPROM (Erasable Programmable Read Only Memory). Examples of EPROM include UV-EPROM (Ultra-Violet Erasable Programmable Read Only Memory), which allows data to be erased by ultraviolet irradiation, EEPROM (Electrically Erasable Programmable Read Only Memory), and flash memory.

[0090] It is preferable that reading comprehension support systems utilize artificial intelligence (AI) for at least some of their processing.

[0091] Reading comprehension support systems preferably utilize artificial neural networks (ANNs, also simply referred to as neural networks). Neural networks are implemented using circuits (hardware) or programs (software).

[0092] In this specification, the term "neural network" refers to any model that mimics the neural network of living organisms, determines the strength of connections between neurons through learning, and possesses problem-solving capabilities. A neural network has an input layer, an intermediate layer (hidden layer), and an output layer.

[0093] In this specification and other documents, when discussing neural networks, the process of determining the connection strength (also called weight coefficient) between neurons from existing information is sometimes referred to as "learning."

[0094] In this specification and other documents, the process of constructing a neural network using connection strengths obtained through learning and deriving new conclusions from it may be referred to as "inference."

[0095] [Output section 140] The output unit 140 outputs information based on the processing results of the processing unit 130. For example, it can supply one or both of the calculation results and / or inference results from the processing unit 130 to an external source of the reading comprehension support system 100. The output unit 140 can also output various data contained in the database based on the processing results of the processing unit 130. The output unit 140 can output information to a display, speaker, etc., used by the user.

[0096] [Transmission path 150] The transmission line 150 has the function of transmitting data. Data can be transmitted and received between the receiving unit 110, the storage unit 120, the processing unit 130, and the output unit 140 via the transmission line 150.

[0097] A reading comprehension support method in a reading comprehension support system according to one embodiment of the present invention will be explained using Figures 2 to 7.

[0098] <Methods to support reading comprehension> One embodiment of the present invention is a reading comprehension support method comprising the processes shown in Figure 2, from step S1 to step S6.

[0099] [Step S1] Step S1 accepts the specified document. The specified document is, for example, a document that the user wants to read. The specified document may be singular or plural.

[0100] Users can directly input text data from a specified document. Alternatively, they may input image data of one or both of the drawings and tables included in the specified document, along with the text data.

[0101] If the data of the specified document is not text data (audio data or image data), the audio data or image data will be converted to text data before proceeding to step S2.

[0102] Furthermore, if the specified document is included in a database, the user can specify the document they wish to read by entering information that identifies the document (searching the database). Based on the information entered by the user, the reading support system retrieves data related to the specified document (specifically, data necessary for subsequent processing) from the database. Information that identifies a document includes the document identification number and the title.

[0103] Furthermore, if a user wants to read a specific part of a document (for example, a particular chapter), they may specify a portion of the document.

[0104] [Step S2] In step S2, a graph representing the structure of the specified document is created using the words and phrases contained within the specified document. If multiple specified documents are specified, a graph is created for each document. In addition, one or more graphs can be created for a single specified document.

[0105] When creating an undirected graph, morphological analysis is first performed on the sentences contained in the specified document. This divides each sentence into tokens, and the part of speech of each token is determined.

[0106] When creating a directed graph, the first step is to perform syntactic analysis on the sentences contained in the specified document. This divides each sentence into tokens, determines the part of speech of each token, and then determines the dependencies between each token.

[0107] In step S2, it is preferable to perform compound word analysis. That is, after the part of speech of a token is determined, it is preferable to generate a new token by combining several tokens. For example, consecutive nouns in a sentence can be combined into one to generate a new token whose part of speech is a compound noun.

[0108] In creating graphs, for example, words that are close together in a document can be directly connected. Each token is used as a label for a node, and each node is connected by an edge. The conditions for connecting nodes with edges can be determined as needed.

[0109] For example, the nodes that connect at an edge can be determined based on the distance between tokens used in the node labels within the document.

[0110] For example, if two phrases exist in the same sentence, those two phrases can be directly joined. Also, for example, if two phrases exist in the same paragraph, those two phrases can be directly joined. Furthermore, for example, if a sentence containing one phrase exists in the vicinity of a sentence containing the other phrase (for example, within n sentences before or after it (where n is an integer of 1 or more, preferably an integer between 1 and 5, more preferably an integer between 3 and 5)), those two phrases can be directly joined.

[0111] Furthermore, in creating a directed graph, the frequency of occurrence and / or importance of each token may be calculated to determine the orientation of the edges.

[0112] In step S2, it is preferable to obtain either or both information regarding the distance between tokens and information regarding the relationship between tokens.

[0113] The distance information of acquired tokens, and the information regarding the relationship between tokens, can be displayed as text labels for edges when visualizing the graph. Alternatively, the color or thickness of the edges may be determined according to the proximity of the tokens. Alternatively, the color or thickness of the edges may be determined according to the strength of the relationship.

[0114] For example, information about the distance between two tokens can be registered as edge information, such as whether the two tokens were in the same sentence, the same paragraph, or how many paragraphs, sentences, words, or strings apart they were.

[0115] For example, information relating to the relationship between two words can be recorded on the edge label, such as whether one word is a related word to the other, and the degree of relevance between the two words. Related words include synonyms, equivalent words, superordinate words, and subordinate words. In addition, other tokens in the sentence that indicate the relationship between the two words (such as noun phrases, verb phrases, and adverbial phrases) can be registered as edge information.

[0116] Figures 3A to 3D illustrate an example of graphing Japanese sentences. Figures 3A to 3D show Japanese text and its corresponding Romanized alphabetical representation.

[0117] Figure 3A shows the statement 300, "The oxide semiconductor layer is above the insulating layer (SANKABUTSUHANDOUTAISOUHAZETSUENTAISOUNOJOUHOUNIARU)".

[0118] In step S2, sentence 300 is subjected to morphological analysis to divide it into multiple tokens, and the part of speech of each token is determined.

[0119] As shown in Figure 3B, sentence 300 is divided into 12 tokens, from token 301 to token 312. In Figure 3B, the part of speech is indicated below each token.

[0120] Next, compound word analysis is performed to combine consecutive nouns into a single word. As a result, sentence 300 is composed of seven tokens, as shown in Figure 3C.

[0121] Specifically, the string of characters for token 301 shown in Figure 3B is "oxidation (SANKA)", the string of characters for token 302 is "thing (BUTSU)", the string of characters for token 303 is "semiconductor (HANDOUTAI)", and the string of characters for token 304 is "layer (SOU)". The part of speech of these tokens 301 through 304 is all nouns. Therefore, as shown in Figure 3C, they can be combined into a single token 321. The string of characters for token 321 is "oxide semiconductor layer (SANKABUTSUHANDOUTAISOU)", and its part of speech is a noun (compound noun).

[0122] Furthermore, the string of characters for token 305 shown in Figures 3B and 3C is "ha (HA)", and its part of speech is a particle.

[0123] Also, the character string of token 306 shown in FIG. 3B is "絶縁 (ZETSUEN)", the character string of token 307 is "体 (TAI)", and the character string of token 308 is "層 (SOU)". The part-of-speech of these tokens 306 to 308 is all noun. Therefore, as shown in FIG. 3C, they are grouped into one token 322. The character string of token 322 is "絶縁体層 (ZETSUENTAISOU)", and the part-of-speech is noun (compound noun).

[0124] Also, the character string of token 309 shown in FIGS. 3B and 3C is "の (NO)", and the part-of-speech is particle. Also, the character string of token 310 is "上方 (JOUHOU)", and the part-of-speech is noun. Also, the character string of token 311 is "に (NI)", and the part-of-speech is particle. Also, the character string of token 312 is "ある (ARU)", and the part-of-speech is verb.

[0125] Next, in step S2, sentence 300 is graphed. FIG. 3D shows an example of graphing sentence 300. Here, an example is shown where tokens 321 and 322 whose part-of-speech is noun are used for the labels of node 323 and node 324, and token 310 whose part-of-speech is noun is used for the edge label 325. Note that in edge label 325, instead of or in addition to the token, at least one of information such as the distance between nodes and information related to the relevance of nodes may be represented.

[0126] The arrow shown in FIG. 3D is illustrated as going from node 323 to node 324. That is, the starting point of the arrow is set as the token that appears earlier in sentence 300, and the ending point of the arrow is set as the token that appears later. Note that the method of determining the direction of the arrow is not limited to this, and the above-described example can be referred to. Therefore, in some cases, the starting point of the arrow may be node 324 and the ending point of the arrow may be node 323. However, it is desirable to unify the method of determining the direction of the arrow within the graph.

[0127] By performing the above process on each sentence in a document, the structure of the entire document can be represented in a single graph. As a result, one or both of nodes 323 and 324 may be further connected to words in other sentences via edges. Furthermore, a portion of the document may be represented in a single graph. Also, a graph may be created for each chapter of the document. In other words, multiple graphs may be created from a single document.

[0128] Figures 4A through 4E illustrate an example of graphing English sentences.

[0129] Figure 4A shows the sentence 330, "A semiconductor device comprising: an oxide semiconductor layer over an insulator layer."

[0130] In step S2, it is preferable to perform a document cleaning process. The cleaning process removes noise contained in the document. For example, this cleaning process may involve deleting semicolons or replacing colons with commas. By performing a cleaning process on the document, the accuracy of morphological analysis can be improved. By performing a cleaning process on sentence 330, the semicolon is deleted, and sentence 330a can be obtained as shown in Figure 4B.

[0131] Next, sentence 330a is divided into multiple tokens by morphological analysis. Although the parts of speech of the tokens are not indicated in Figure 4C, the part of speech of each token can be determined through morphological analysis.

[0132] As shown in Figure 4C, sentence 330a is divided into 12 tokens, from token 331 to token 342.

[0133] Next, compound word analysis is performed to combine consecutive nouns into one. As a result, sentence 330a consists of five tokens, as shown in Figure 4D.

[0134] Specifically, the string of token 331 shown in Figure 4C is "A", the string of token 332 is "semiconductor", and the string of token 333 is "device". The part of speech of token 331 is an indefinite article, and the parts of speech of tokens 332 and 333 are both nouns. Therefore, as shown in Figure 4D, they can be combined into a single token 351. The string of token 351 is "A semiconductor device", and its part of speech is a noun (compound noun).

[0135] Furthermore, the string of token 334 shown in Figures 4C and 4D is "comprising".

[0136] Furthermore, the string of token 335 shown in Figure 4C is "an", the string of token 336 is "oxide", the string of token 337 is "semiconductor", and the string of token 338 is "layer". The part of speech of token 335 is an indefinite article, and the part of speech of tokens 336 through 338 is all nouns. Therefore, as shown in Figure 4D, they can be combined into a single token 352. The string of token 352 is "an oxide semiconductor layer", and its part of speech is a noun (compound noun).

[0137] Furthermore, the string of token 339 shown in Figures 4C and 4D is "over".

[0138] Furthermore, the string of token 340 shown in Figure 4C is "an", the string of token 341 is "insulator", and the string of token 342 is "layer". The part of speech of token 340 is an indefinite article, and the parts of speech of tokens 341 and 342 are both nouns. Therefore, as shown in Figure 4D, they can be combined into a single token 353. The string of token 353 is "an insulator layer", and its part of speech is a noun (compound noun).

[0139] Next, in step S2, sentence 330 is graphed. Figure 4E shows an example of a graph of sentence 330. Here, tokens 351 to 353, whose part of speech is noun, are used as labels for nodes 354 to 356, token 334 is used as the label 357 for the edge between node 354 and node 355, and token 339 is used as the label 358 for the edge between node 355 and node 356.

[0140] One of the arrows shown in Figure 4E is illustrated to point from node 354 to node 355, and the other arrow is illustrated to point from node 355 to node 356. In other words, the starting point of the arrow is the token that appears earlier in sentence 330, and the ending point of the arrow is the token that appears later.

[0141] In this embodiment, the process of creating a graph from a document was explained using examples of documents written in Japanese and documents written in English, but there are no particular limitations on the language of the document. For example, graphs can be created from documents written in languages ​​such as Chinese, Korean, German, French, Russian, and Hindi by following the same process.

[0142] [Step S3] Step S3 outputs multiple words and phrases included in the graph.

[0143] There are no particular restrictions on the output method; for example, a list of words can be displayed as a list. Alternatively, the graph created in step S2 can be displayed itself. Or, both the graph and the list can be displayed.

[0144] [Step S4] Step S4 accepts multiple specified terms.

[0145] The user selects multiple specified terms from the multiple terms output in step S3.

[0146] Table 1 shows an example where multiple terms are displayed as a list in step S3, and the user specifies the terms in step S4. As shown in Table 1, the following explanation will use the example where two terms, "layer A" and "layer B," are selected as the multiple specified terms.

[0147] [Table 1]

[0148] [Step S5] In step S5, the graph is searched using the multiple specified keywords received in step S4.

[0149] Specifically, in step S5, the shortest path between each of the specified terms in the graph can be calculated.

[0150] Figure 5A shows an example of the graph created in step S2, with only the parts related to "layer A" and "layer B" extracted.

[0151] The graph shown in Figure 5A has nodes 151 through 156. "layer A" is the label for node 151, and "layer B" is the label for node 152. In addition, node 153, which has "layer C" as its label, node 154, which has "word D" as its label, node 155, which has "word E" as its label, and node 156, which has "word F" as its label, are included in the path connecting node 151 and node 152.

[0152] In Figures 5 through 7, nodes to which the specified term has been assigned as a label are indicated by diagonal hatching.

[0153] If the cost required to traverse each edge is the same (i.e., all edge weights are the same), then the shortest path is the one that involves the fewest nodes. In other words, in the graph shown in Figure 5A, the shortest path connecting node 151 and node 152 is the path that goes through node 153, which has "layer C" as its label (the path shown as a thick line in Figure 5A). In this way, the shortest path between each of the multiple specified terms is calculated.

[0154] [Step S6] In step S6, the results of the graph search in step S5 are output.

[0155] Figure 5B shows the shortest path connecting node 151 and node 152 in Figure 5A. By outputting the graph shown in Figure 5B, the relationship between “layer A” and “layer B” can be presented. In Figure 5B, since “layer A” and “layer B” are connected via the complementary term “layer C,” which is a different term from the specified term, it can be shown to the user that “layer C” may be strongly related to the information the user wants to understand.

[0156] Furthermore, information about multiple specified terms can be presented using at least one of the edge labels, orientation, color, and thickness.

[0157] In Figure 5C, the undirected graph shown in Figure 5B is represented as a directed graph. Additionally, the edge between node 151 and node 153 is labeled 159, and the edge between node 153 and node 152 is labeled 160.

[0158] From label 159 shown in Figure 5C, we can see that “layer A” is a superordinate term to “layer C”. A specific example of “layer A” is a “semiconductor layer”, and a specific example of “layer C” is an “oxide semiconductor layer”.

[0159] Furthermore, the label 160 contains the word "over," indicating that "layer C" is located above "layer B." In this way, edge information can be used to present users with information about the specified words displayed in the nodes.

[0160] Even if the graph search results in step S5 are the same, the graph displayed in step S6 is not necessarily unique. For example, the edge lengths and the corresponding node positions can be displayed in several different ways and are not particularly limited.

[0161] Furthermore, it is preferable to extract and output sentences from the document based on each shortest path.

[0162] Figure 6 shows an example of the output. In Figure 6, the three specified terms are “layer A”, “layer B”, and “device G”.

[0163] Graph 510, shown in Figure 6, has nodes 151 through 153, node 157, and node 158. "layer A" is the label for node 151, "layer B" is the label for node 152, and "device G" is the label for node 157. In addition, node 153, which has "layer C" as its label, and node 158, which has "word H" as its label, are also included in graph 510.

[0164] Graph 510 shows the shortest paths between each of the multiple specified terms. It can be seen that the shortest path between "layer A" and "layer B" is via the complementary term "layer C". It can be seen that the shortest path between "layer A" and "device G" is a direct path. It can be seen that the shortest path between "device G" and "layer B" is via the complementary term "word H".

[0165] The extracted text 520 shown in Figure 6 is the result of extracting sentences from the document based on each shortest path. Here, we will explain using the example where graph 510 was created by directly connecting tokens contained in the same sentence or paragraph.

[0166] From the extracted text 520, we can see that "layer A" and "layer C" are included in the same sentence in the 10th paragraph, and "layer C" and "layer B" are included in the same sentence in the 15th paragraph. In this way, by extracting descriptions related to specified terms, users can efficiently read and understand the document even when the sentences are located far apart. Note that in the extracted text 520, only the sentences containing the specified terms from each paragraph may be displayed, or the entire paragraph containing the specified terms may be displayed.

[0167] From the extracted text 520, we can see that "layer A" and "device G" are included in the same sentence in paragraph 30. Note that the order in which specified words appear in the sentence does not matter when extraction is performed.

[0168] From the extracted text 520, we can see that "layer B" and "word H" are in the same sentence in paragraph 16. Furthermore, we can see that "word H" and "device G" are in different sentences in paragraph 38. Thus, even when two terms are in different sentences, extracting both sentences when they are in the same paragraph can provide more detailed information about the specified terms. By extracting descriptions related to the specified terms, users can efficiently read and understand the document, even when the sentences are located far apart.

[0169] Furthermore, if the extracted text 520 contains information such as figures, tables, mathematical formulas, or chemical formulas, it is preferable to display images of such figures, tables, mathematical formulas, or chemical formulas together with the text. This can better assist the user in understanding the document. For example, it is preferable to display "Fig. X" and "Table Z" shown in Figure 6, or links to these figures and tables, together with the graph 510 and the extracted text 520.

[0170] Figure 7 shows an example of a graph output different from that in Figure 6.

[0171] Figure 7 shows an example where five terms were selected as specified terms: “layer A”, “layer B”, “layer C”, “layer D”, and “layer E”.

[0172] The graph shown in Figure 7 has nodes 161 through 167. "layer A" is the label for node 161, "layer B" is the label for node 162, "layer C" is the label for node 163, "layer D" is the label for node 164, and "layer E" is node 165. In addition, node 166, which has "word X" as its label, and node 167, which has "word Y" as its label, are also included in the graph.

[0173] Figure 7 shows the shortest paths between each specified word. For example, it can be seen that the shortest path between "layer A" and "layer B" is a direct connection. Similarly, it can be seen that the shortest path between "layer A" and "layer C" is a direct connection. It can also be seen that the shortest path between "layer A" and "layer E" is a connection via the complementary word "word Y".

[0174] Figure 7 shows that node 164 is not connected to any other nodes. This suggests that the specified document may lack sufficient information or contain errors regarding "layer D".

[0175] In this way, the results of a graph search can also be used to proofread a document.

[0176] Furthermore, it can be seen that there are two shortest paths connecting “layer B” and “layer E”: one through the specified term “layer C” and the complementary term “word Y”, and another through the complementary terms “word X” and “word Y”. In this case, the two shortest paths are shown, and sentences can be extracted based on each.

[0177] Furthermore, even when multiple documents are specified, the graph can be created and the search performed in the same manner as described above, and the search results can be output. By comparing the output results, the user can easily confirm the relationships and differences between multiple documents.

[0178] Furthermore, the graph showing the shortest path between specified terms, which is the search result, can be vectorized, and the similarity of each vector can be calculated to evaluate the similarity of multiple documents and present it to the user.

[0179] As described above, the reading comprehension support system of this embodiment can provide the user with reading comprehension support by presenting a graph showing the relationships between multiple specified words in a document specified by the user. By using the system to extract and output sentences containing multiple specified words, the user can efficiently read through the document. This allows the user to quickly find the necessary information from the document.

[0180] This embodiment can be combined with other embodiments as appropriate. Furthermore, if multiple configuration examples are shown within a single embodiment in this specification, these configuration examples can be combined as appropriate.

[0181] (Embodiment 2) In this embodiment, a reading comprehension support system according to one aspect of the present invention will be described with reference to Figures 8 and 9.

[0182] <Reading Comprehension Support System 2> Figure 8 shows a block diagram of the reading comprehension support system 210. The reading comprehension support system 210 includes a server 220 and a terminal 230 (such as a personal computer). For the same components as the reading comprehension support system 100 shown in Figure 1, please also refer to the description of <Reading Comprehension Support System 1> in Embodiment 1.

[0183] The server 220 includes a communication unit 171a, a transmission line 172, a storage unit 120, and a processing unit 130. Although not shown in Figure 8, the server 220 may also have at least one of the following: a reception unit, a database, an output unit, an input unit, etc.

[0184] Terminal 230 includes a communication unit 171b, a transmission line 174, an input unit 115, a storage unit 125, a processing unit 135, and a display unit 145. Examples of terminal 230 include tablet personal computers, notebook personal computers, and various portable information terminals. Alternatively, terminal 230 may be a desktop personal computer without a display unit 145, and may be connected to a monitor or the like that functions as a display unit 145.

[0185] The user of the reading comprehension support system 210 inputs information about a specified document from the input unit 115 of the terminal 230 to the server 220. This information is then transmitted from the communication unit 171b to the communication unit 171a.

[0186] For example, text data of a specified document is transmitted from communication unit 171b to communication unit 171a. In addition, image data of at least one of the following may be transmitted: drawings, chemical formulas, mathematical formulas, and tables. Also, for example, information identifying the document is transmitted from communication unit 171b to communication unit 171a.

[0187] The information received by the communication unit 171a is stored in the memory or storage unit 120 of the processing unit 130 via the transmission line 172. Alternatively, information may be supplied from the communication unit 171a to the processing unit 130 via the receiving unit (see receiving unit 110 shown in Figure 1).

[0188] The various processes described in the <Reading Comprehension Support Method> of Embodiment 1 are performed in the processing unit 130. Since these processes require high processing power, it is preferable that they be performed in the processing unit 130 of the server 220. It is preferable that the processing unit 130 has higher processing power than the processing unit 135.

[0189] The processing result of the processing unit 130 is stored in the memory or storage unit 120 of the processing unit 130 via the transmission line 172. Subsequently, the processing result is output from the server 220 to the display unit 145 of the terminal 230. The processing result is transmitted from the communication unit 171a to the communication unit 171b. In addition, various data contained in the database may be transmitted from the communication unit 171a to the communication unit 171b based on the processing result of the processing unit 130. Furthermore, the processing result may be supplied from the processing unit 130 to the communication unit 171a via the output unit (output unit 140 shown in Figure 1).

[0190] [Communication section 171a and communication section 171b] Data can be sent and received between the server 220 and the terminal 230 using the communication units 171a and 171b. The communication units 171a and 171b can be hubs, routers, modems, etc. Data can be sent and received using either wired or wireless (e.g., radio waves, infrared, etc.) connections.

[0191] [Transmission lines 172 and 174] Transmission lines 172 and 174 have the function of transmitting data. Data can be transmitted and received between the communication unit 171a, the storage unit 120, and the processing unit 130 via transmission line 172. Data can be transmitted and received between the communication unit 171b, the input unit 115, the storage unit 125, the processing unit 135, and the output unit 140 via transmission line 174.

[0192] [Input section 115] The input unit 115 can be used by the user to specify documents and phrases. For example, the input unit 115 may have functions to operate the terminal 230, specifically including a mouse, keyboard, touch panel, microphone, scanner, camera, etc.

[0193] The reading comprehension support system 210 may have a function to convert audio data into text data. For example, at least one of the processing unit 130 and the processing unit 135 may have such a function.

[0194] The reading comprehension support system 210 may have an optical character recognition (OCR) function. This allows it to recognize characters contained in image data and create text data. For example, at least one of the processing unit 130 and processing unit 135 may have this function.

[0195] [Storage section 125] The storage unit 125 may store either or both data relating to the specified document and / or data supplied from the server 220. Furthermore, the storage unit 125 may hold at least some of the data that the storage unit 120 may hold.

[0196] [Processing unit 130 and processing unit 135] The processing unit 135 has the function of performing calculations and other operations using data supplied from the communication unit 171b, the storage unit 125, and the input unit 115, etc. The processing unit 135 may also have the function of performing at least a part of the processing that can be performed by the processing unit 130.

[0197] The processing unit 130 and the processing unit 135 may each have either or both of the following: a transistor having a metal oxide in its channel formation region (OS transistor), and a transistor having silicon in its channel formation region (Si transistor).

[0198] In this specification, a transistor using an oxide semiconductor or metal oxide in the channel formation region is referred to as an oxide semiconductor transistor or OS transistor. The channel formation region of an OS transistor preferably contains a metal oxide.

[0199] In this specification, "metal oxide" refers to an oxide of a metal in a broad sense. Metal oxides are classified into oxide insulators, oxide conductors (including transparent oxide conductors), oxide semiconductors (also called oxide semiconductors or simply OS), etc. For example, when a metal oxide is used in the semiconductor layer of a transistor, the metal oxide may be referred to as an oxide semiconductor. In other words, if a metal oxide has at least one of the following properties: amplification, rectification, and switching, the metal oxide can be called a metal oxide semiconductor, or OS for short.

[0200] The metal oxide in the channel-forming region preferably contains indium (In). When the metal oxide in the channel-forming region contains indium, the carrier mobility (electron mobility) of the OS transistor increases. Furthermore, the metal oxide in the channel-forming region is preferably an oxide semiconductor containing element M. Element M is preferably at least one of aluminum (Al), gallium (Ga), and tin (Sn). Other elements that can be used for element M include boron (B), silicon (Si), titanium (Ti), iron (Fe), nickel (Ni), germanium (Ge), yttrium (Y), zirconium (Zr), molybdenum (Mo), lanthanum (La), cerium (Ce), neodymium (Nd), hafnium (Hf), tantalum (Ta), and tungsten (W). However, it is also possible to combine multiple of the above elements as element M. Element M is, for example, an element with a high bond energy with oxygen. For example, the element has a higher bonding energy with oxygen than indium. Furthermore, the metal oxide containing the channel-forming region is preferably a zinc (Zn)-containing metal oxide. Zinc-containing metal oxides may be more prone to crystallization.

[0201] The metal oxide present in the channel-forming region is not limited to indium-containing metal oxides. The semiconductor layer may be, for example, zinc-tin oxide, gallium-tin oxide, or other metal oxides that do not contain indium, such as zinc-containing metal oxides, gallium-containing metal oxides, or tin-containing metal oxides.

[0202] The processing unit 130 preferably has an OS transistor. Because the OS transistor has an extremely small off-current, using the OS transistor as a switch to hold the charge (data) that has flowed into a capacitive element that functions as a memory element ensures that the data can be retained for a long period of time. By using this characteristic in at least one of the registers and cache memory of the processing unit 130, the processing unit 130 can be operated only when necessary, and the information of the previous processing can be saved to the memory element in other cases, thereby turning off the processing unit 130. In other words, normally-off computing becomes possible, and the power consumption of the reading support system can be reduced.

[0203] [Display section 145] The display unit 145 has the function of displaying the output result. Examples of the display unit 145 include liquid crystal displays and light-emitting displays. Examples of light-emitting elements that can be used in light-emitting displays include LEDs (Light Emitting Diodes), OLEDs (Organic LEDs), QLEDs (Quantum-dot LEDs), and semiconductor lasers. In addition, the display unit 145 can also use a display device using a shutter-type or optical interference-type MEMS (Micro Electro Mechanical Systems) element, a display device using a display element that applies a microcapsule type, electrophoretic type, electrowetting type, or electronic powder fluid (registered trademark) type.

[0204] Figure 9 shows an illustrative diagram of the reading comprehension support system of this embodiment.

[0205] The reading comprehension support system shown in Figure 9 comprises a server 5100 and terminals (which can also be called electronic devices). Communication between the server 5100 and each terminal can be performed via an internet connection 5110.

[0206] The server 5100 can perform calculations using data input from the terminal via the internet connection 5110. The server 5100 can also transmit the results of the calculations to the terminal via the internet connection 5110. This reduces the computational burden on the terminal.

[0207] Figure 9 shows information terminals 5300, 5400, and 5500 as terminals. Information terminal 5300 is an example of a mobile information terminal such as a smartphone. Information terminal 5400 is an example of a tablet terminal. In addition, information terminal 5400 can be used as a notebook-type information terminal by connecting it to a chassis 5450 that has a keyboard. Information terminal 5500 is an example of a desktop-type information terminal.

[0208] By configuring the system in this way, users can access the server 5100 from information terminals 5300, 5400, and 5500, etc. Users can then receive services provided by the administrator of the server 5100 through communication via the internet line 5110. Examples of such services include a service using a reading comprehension support method according to one aspect of the present invention. In such a service, artificial intelligence may be used on the server 5100.

[0209] This embodiment can be combined with other embodiments as appropriate. [Explanation of symbols]

[0210] 100: Reading comprehension support system, 110: Reception unit, 115: Input unit, 120: Memory unit, 125: Memory unit, 130: Processing unit, 135: Processing unit, 140: Output unit, 145: Display unit, 150: Transmission line, 151: Node, 152: Node, 153: Node, 154: Node, 155: Node, 156: Node, 157: Node, 158: Node, 159: Label, 160: Label, 161: Node, 162 :Node, 163:Node, 164:Node, 165:Node, 166:Node, 167:Node, 171a:Communication Unit, 171b:Communication Unit, 172:Transmission Line, 174:Transmission Line, 210:Reading Comprehension Support System, 220:Server, 230:Terminal, 300:Text, 301:Token, 302:Token, 303:Token, 304:Token, 305:Token, 306:Token, 307:Token, 308: Token, 309: Token, 310: Token, 311: Token, 312: Token, 321: Token, 322: Token, 323: Node, 324: Node, 325: Label, 330a: Sentence, 330: Sentence, 331: Token, 332: Token, 333: Token, 334: Token, 335: Token, 336: Token, 337: Token, 338: Token, 339: Token, 340: Token, 341: Token, 342: Token, 351: Token, 352: Token, 353: Token, 354: Node, 355: Node, 356: Node, 357: Label, 358: Label, 510: Graph, 520: Extracted text, 5100: Server, 5110: Internet connection, 5300: Information terminal, 5400: Information terminal, 5450: Enclosure, 5500: Information terminal

Claims

1. A reading comprehension support method using a reading comprehension support system in which graphs are stored, The aforementioned reading comprehension support system is We accept the specified documents. Using the words and phrases contained in the aforementioned designated document, a first graph representing the structure of the aforementioned designated document is created. Output two or more words included in the first graph, The system accepts multiple specified phrases from the outputted phrases. A reading comprehension support method that searches the first graph using the aforementioned multiple specified terms and outputs the search results.

2. In claim 1, A reading comprehension support method that, as a result of the search, outputs a second graph showing at least the shortest path between any two of the plurality of specified words in the first graph.

3. In claim 2, A reading comprehension support method that outputs, along with the search results, sentences containing the specified words in paragraphs within the specified document that contain two or more of the specified words.

4. In claim 2 or 3, The shortest path is a path that connects any two of the plurality of specified terms via at least one complementary term. A reading comprehension support method in which the aforementioned supplementary phrase is a phrase different from the aforementioned multiple specified phrases.

5. In claim 4, A reading comprehension support method that outputs, along with the search results, a sentence in a paragraph of the specified document that contains at least one of the specified words and at least one of the supplementary words, wherein the paragraph contains at least one of the specified words and the supplementary word.

6. In claim 1, A reading comprehension support method that, as a result of the search, outputs a second graph showing at least the shortest path between each of the plurality of specified words in the first graph.

7. In claim 6, A reading comprehension support method that outputs, along with the search results, sentences containing the specified words in paragraphs within the specified document that contain two or more of the specified words.

8. In claim 6 or 7, The shortest path connecting any two of the aforementioned specified terms is a path connecting the two specified terms via at least one complementary term. A reading comprehension support method in which the aforementioned supplementary phrase is a phrase different from the aforementioned multiple specified phrases.

9. In claim 8, A reading comprehension support method that outputs, along with the search results, a sentence in a paragraph of the specified document that contains at least one of the specified words and at least one of the supplementary words, wherein the paragraph contains at least one of the specified words and the supplementary word.

Images