KR101417928B1 - Method for Structuring Natural Language And Mathematical Formula, Apparatus And Computer-Readable Recording Medium with Program Therefor - Google Patents

Abstract

An embodiment of the present invention relates to a method for structuring natural language and mathematical expressions, an apparatus therefor, and a computer-readable recording medium.
According to an embodiment of the present invention, there is provided an information processing apparatus comprising: an information input unit for inputting combination data composed of a combination of a natural language and a mathematical formula; A separator for separating the natural language and the expression from the combination data; A natural language processing unit for analyzing each first information constituting the separated natural language and classifying the first information according to a specific meaning; A mathematical expression processing unit for analyzing each of the second information constituting the separated mathematical expression and classifying the second information according to a specific meaning; And a data management unit for re-assembling at least one of the first information, the second information, the natural language, and the formula, and storing the recombined data as the recombination data.
According to an embodiment of the present invention, there is an effect that a keyword can be used as a keyword when searching for math contents using a standard document in which natural language and mathematical expression are combined together.

Description

TECHNICAL FIELD The present invention relates to a method for structuring a natural language and a formula, an apparatus therefor, and a computer-readable recording medium,

An embodiment of the present invention relates to a method for structuring natural language and mathematical expressions, an apparatus therefor, and a computer-readable recording medium. More specifically, since it is difficult to analyze data in which a combination of a natural word and a mathematical expression is analyzed by a general natural language processing technique, data obtained by recombining natural language and mathematical expression based on the analysis contents in which the natural language processing A method for structuring a natural language and a formula, an apparatus therefor, and a computer readable recording medium.

The contents described in this section merely provide background information on the embodiment of the present invention and do not constitute the prior art.

The language of a person is rich and complex, and includes enormous vocabularies with complex grammatical and contextual meanings, but machine or software applications generally require data to be entered according to a particular format or rule. Here, natural language input can be used in almost all software applications for interacting with people. A typical natural language processing process separates the natural language into tokens, maps to one or more operations provided by the software application, and each software application is set to have a set of unique operations. That is, software developers write code that interprets natural language input and maps the input to the appropriate behavior for each application.

However, there is a problem that the natural language processing method does not recognize the formula.

According to an embodiment of the present invention, there is provided a method of managing natural language and mathematical expression data by combining natural language and mathematical expression based on analysis results obtained by performing natural language processing and mathematical expression processing on data combined with natural language and mathematical expression A method for structuring natural language and expressions, an apparatus therefor, and a computer-readable recording medium.

According to an aspect of the present invention, there is provided an information processing apparatus comprising: an information input unit for receiving combination data composed of a combination of a natural language and a mathematical formula; A separator for separating the natural language and the expression from the combination data; A natural language processing unit for analyzing each first information constituting the separated natural language and classifying the first information according to a specific meaning; A mathematical expression processing unit for analyzing each of the second information constituting the separated mathematical expression and classifying the second information according to a specific meaning; And a data management unit for re-assembling at least one of the first information, the second information, the natural language, and the formula, and storing the recombined data as the recombination data.

According to another aspect of the present invention, there is provided an information processing apparatus comprising: an information input unit for receiving combination data composed of a combination of a natural language and a mathematical formula; A separator for separating the natural language and the expression from the combination data; Generates stop word filtering data in which a stop word is filtered based on the natural language token, generates a stop word filtering data by filtering the stop word filtering data, A natural language processing unit for generating the de-duplication filtering data and matching the de-duplication filtering data with the pre-defined meaningful operation information; A mathematical expression processing unit for analyzing each of the second information constituting the separated mathematical expression and classifying the second information according to a specific meaning; And a data management unit configured to recombine at least one of the natural language token, the operation information, the second information, the natural language, and the mathematical expression, and store the recombined data as the recombination data.

According to another aspect of the present invention, there is provided an information processing apparatus comprising: an information input unit for receiving combination data composed of a combination of a natural language and a mathematical formula; A separator for separating the natural language and the expression from the combination data; A natural language processing unit for analyzing each first information constituting the separated natural language and classifying the first information according to a specific meaning; A mathematical expression processing unit for converting the separated expression into a tree form, performing a traverse process on the expression converted into the tree form, and generating a mathematical token in which the tokenization is performed on the mathematical expression in which the traversal process is performed; And a data management unit for recombining at least one of the first information, the expression token, the natural language, and the expression, and storing the recombined data as the recombination data.

According to another aspect of the present invention, there is provided an information input method comprising: inputting combination data composed of a combination of a natural language and a mathematical formula; A separating step of separating the natural language and the expression from the combination data, respectively; A natural language processing step of analyzing each first information constituting the separated natural language and classifying the first information according to a specific meaning; A mathematical expression processing step of analyzing each of the second information constituting the separated mathematical expression and classifying it according to a specific meaning; And a data management step of re-assembling at least one of the first information, the second information, the natural language, and the formula, and storing the recombined data as the recombination data.

According to another aspect of the present invention, there is provided a computer-readable recording medium storing a program for executing each step of a natural language and a method for structuring a mathematical expression.

As described above, according to an embodiment of the present invention, it is possible to manage natural language and mathematical expression data by recombining data based on analysis contents in which natural language processing and mathematical expression processing are performed on data combined with natural language and mathematical expression It is effective.

In addition, according to an embodiment of the present invention, information obtained by analyzing natural language and mathematical expression is managed as a recombined data and converted into a mathematical standard document. In addition, according to an embodiment of the present invention, there is an effect that a standard document combining natural language and mathematical expression together can be used as a keyword in searching for math contents in the future.

According to an embodiment of the present invention, semantic information obtained through a natural language can be defined as a natural language token, thereby grasping a configuration pattern of a natural language and grasping an operation described in a natural language. The semantic information obtained through the above-described method can be defined as a Mathematics Natural Language token, and the data can be managed in the form of data including natural language and mathematical expressions.

1 is a block diagram schematically showing an apparatus for structuring natural language and mathematical expressions according to an embodiment of the present invention,
2 is a block diagram schematically showing a natural language processing unit according to an embodiment of the present invention,
3 is a block diagram schematically showing a modification processing unit according to an embodiment of the present invention.
4 is a flowchart for explaining a method of structuring a natural language and an equation according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating a tree form representation of a formula according to an embodiment of the present invention. FIG.
6 is a diagram illustrating an example of a system in which a natural language and expression structuring apparatus according to an embodiment of the present invention provides data in a cloud computing;
7 is a diagram illustrating an example of a method of analyzing information constituting a natural language and a formula according to an embodiment of the present invention and classifying the information according to a specific meaning.

The natural language and mathematical expression structuring apparatus 100 according to the present invention refers to an apparatus for structuring (DB) by natural language and mathematical expression in combination data composed of a combination of a natural language and a mathematical formula, The structuring apparatus 100 may be implemented by hardware or software, and may be mounted on a server or a terminal.

1 is a block diagram schematically showing a natural language and expression structuring apparatus according to an embodiment of the present invention.

The natural language and mathematical structuring apparatus 100 according to an embodiment of the present invention includes an information input unit 110, a separating unit 120, a natural language processing unit 130, a mathematical expression processing unit 140, and a data management unit 150. In an embodiment of the present invention, the natural language and mathematical structuring apparatus 100 includes only the information input unit 110, the separating unit 120, the natural language processing unit 130, the modification processing unit 140, and the data management unit 150 It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the appended claims. Various modifications and changes may be made to the elements included in the natural language and mathematical expression structuring apparatus 100 within the scope of the present invention.

The information input unit 110 receives combination data composed of combinations of natural language and mathematical expressions. Here, the combination data is preferably mathematical content including mathematical problems, proofs of formulas, but is not limited thereto. In addition, the combination data composed of a combination of natural language and mathematical expression can be directly input by a user's operation or command, but not necessarily limited thereto. Document data composed of natural language and combination of mathematical expressions may be input from a separate external server will be. The separator 120 separates the natural language and the expression from the combination data, respectively. That is, when the combination data composed of a combination of natural language and mathematical expression is input through the information input unit 110, the separation unit 120 separates and recognizes the natural language and the mathematical expression included in the combination data.

The natural language processing unit 130 analyzes the first information constituting the separated natural language and classifies the first information according to a specific meaning. The natural language processing unit 130 analyzes the first information constituting the natural language and then analyzes the structure of the sentence and the keyword included in the keyword A specific meaning can be grasped by using at least one information. That is, the natural language processing unit 130 operates on the basis of a predetermined rule and can grasp a specific meaning. The natural language processing unit 130 analyzes a first information constituting a natural language and divides the information according to a specific meaning Will be described with reference to FIG.

The natural language processing unit 130 generates a natural language token obtained by tokenizing a natural language. Token refers to a unit that can be distinguished in a continuous sentence. Tokenization refers to a process of dividing a natural language into units of words (Word), which is a unit understood by the natural language and mathematical structuring apparatus 100. More specifically tokenization, in one embodiment of the present invention tokenization is largely divided into natural language tokenization and formal tokenization. Natural tokenization is a process of recognizing each word (Word) corresponding to the result obtained by separating natural words included in the combination data (mathematical problem) on the basis of a space as a natural language token. In order to more clearly understand the meaning of each token, an additional morphological analysis of the token may be performed. Meanwhile, the formula tokenization process refers to a process of recognizing individual unit information obtained after parsing a formula included in a combination data (mathematical problem) as a formula token.

[Example 1]

For example, in [Example 1], the information corresponding to the natural language token is 'Find', 'the', 'function', 'value', 'with', and the formula token is returned after parsing (Polynomial), Max degree = 3, Number of terms = 4, Condition, and so on.

The natural language processing unit 130 generates word filtering data by filtering the stop word based on the natural language token and generates the deduplication filtering data by performing the deduplication filtering on the stop word filtering data. Here, a stop word refers to a set of words that are predefined in order to remove a part corresponding to a token which is unnecessary in the analysis of a sentence or an expression. That is, in [Example 1] 'the' (other than a or to) is predefined as a dictionary in the system. Here, the dictionary means a list including a set of words. That is, after generating the natural language token, the natural language processing unit 130 performs a process of removing a stop word, which is a part unnecessary in the analysis. The stop word filtering is performed in a case where a mathematical problem is prolonged It prevents many tokens from entering and also works to improve the processing speed of the system.

The natural language processing unit 130 matches the action information assigned with the predefined meaning to the deduplication filtering data. Here, the operation information means summary information that can be extracted based on a natural language token or an expression token. For example, in [Example 1], action information called 'Solve' can be extracted based on a natural language token or a mathematical token. Here, the reason why the data corresponding to the predicate in the deduplication filtering data is matched with the operation information is that in the process of defining the combination data (mathematical problem) as a schema, information on the representative operation of the whole sentence is obtained This is to be used later as a tool to assist in analyzing the similarity between searches or problems.

The natural language processing unit 130 generates a natural language token by performing tokenization on the first information constituting the natural language. The natural language processing unit 130 generates stop word filtering data by performing stop word filtering for selecting and removing the natural language tokens identified as predetermined stop words in the natural language token. The natural language processing unit 130 performs deduplication filtering to selectively remove redundant data from the stop word filtering data to generate deduplication filtering data. The natural language processing unit 130 stores the data corresponding to the predicate in the deduplication filtering data and stores the matching operation information with the predefined meaning.

The mathematical expression processing unit 140 analyzes each of the second information constituting the separated mathematical expression and classifies it according to a specific meaning. To describe the operation performed by the expression processor 140 in order to grasp a specific meaning, the expression processor 140 analyzes the second information constituting the expression, and then, using the expression information of the expression, . That is, the mathematical expression processing unit 140 can operate on the basis of a predetermined rule to grasp a specific meaning, and the mathematical expression processing unit 140 analyzes the second information constituting the mathematical expression, Will be described with reference to FIG.

The mathematical expression processing unit 140 converts a mathematical expression into a tree form, performs a traverse process on a mathematical expression transformed into a tree form, and performs tokenization on a mathematical expression in which a traversal process is performed. The mathematical expression processing unit 140 converts mathematical expressions written in Math ML (Mathematical Markup Language) into an XML tree form, and then converts it into a DOM (Document Object Model) form. The mathematical expression processing unit 140 executes the traversal in a depth-first search method in which the second information constituting the equation is gradually transferred from the lowermost node to the upper node. On the other hand, to describe the transversal process and the depth-first search in detail, in general, the formula is in the form of a Math ML, which is configured in the form of a tree. Such a process of traversing a tree is called a traversal process, When performing the process, use Depth-First Search. The traversal process starts from the root of the tree and moves to the child node. When all the child nodes are searched, the child node moves to the parent node. Therefore, all the information held by the child node is transferred to the parent node. In terms of time complexity, it is efficient to search only the number of edges.

The data management unit 150 stores at least one or more of the first information analyzed through the natural language processing unit 130, the second information analyzed through the modification processing unit 140, the natural language recognized through the separation unit 120, Recombine and store as recombination data. The data management unit 150 converts the recombined data into document data. Meanwhile, the data management unit 150 can define XML so that the first information, the second information, the natural language, and the formula are stored in one XML (eXtended Markup Language) tree. In the present embodiment, however, do. However, to briefly describe the XML in which the first information, the second information, the natural language, and the formula are defined, the defined XML can be roughly divided into two parts. First, the 'problem description' The second part can be divided into 'semantic' part based on information extracted from natural language and formula. Here, the 'semantic' part may be added or changed in the future depending on the discovery of new mathematical problem types.

In addition, the mathematical problem is defined as XML. The mathematical problem is organized in a tree form, and information necessary for the mathematical problem is gathered into a semantic part in the whole tree. Indexing) or the like. That is, according to the mathematical problem configured in the tree form, the mathematical contents expressed by the natural language and the standardized mathematical expression are converted into the form that can be understood (recognized) by the natural language and mathematical formula structuring apparatus 100, The semantic information can be extracted and structured in an XML tree form.

Meanwhile, the natural language and mathematical structuring apparatus 100 stores computing resources such as hardware or software for structuring natural language and mathematical expressions, and provides the computing resources required by the client to the corresponding terminals by means of cloud computing There will be. A detailed description related to this will be given in FIG.

2 is a block diagram schematically showing a natural language processing unit according to an embodiment of the present invention.

The natural language processing unit 130 according to an exemplary embodiment of the present invention includes a natural language tokenizing unit 210, a stop word filtering unit 220, a deduplication filtering unit 230, and an operation matching unit 240. In an embodiment of the present invention, the natural language processing unit 130 includes only the natural language tokenizing unit 210, the stop word filtering unit 220, the deduplication filtering unit 230, and the operation matching unit 240 It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims. Various modifications and changes may be made to the components included in the natural language processing unit 130. [

The natural language tokenizing unit 210 generates a natural language token by tokenizing the natural language. The natural language tokenizing unit 210 generates a natural language token by performing tokenization on the first information constituting the natural language. Here, a natural language token refers to each word corresponding to a result obtained by separating the natural language included in the combination data (mathematical problem) on the basis of a space. For example, the natural language and mathematical structuring apparatus 100 may receive the natural language nodes included in the combination data individually or all of the natural language nodes at once by using the natural language tokenizing unit 210. Here, the natural language has a nature of a sentence composed of a plurality of words, but is not limited to a complete sentence. That is, the natural language node is divided into words, which is a unit understood by the natural language and mathematical formula structuring apparatus 100, and this process is called a tokenizing process. When constructing a combination data (mathematical problem) with a schema, the natural language and the mathematical expression form a mixed form irrespective of the order. At this time, the natural language node is referred to as a natural language node. That is, a plurality of natural language parts may be included in one problem (i.e., schema). [Example 1] contains two natural language nodes, 'Find the function value' and 'with' are natural language nodes. Therefore, when inputting to a system, a tokenization process is performed to divide a natural language node into units that can be understood by the system. Here, a natural language token refers to each word corresponding to a result obtained by separating the natural language included in the combination data (mathematical problem) on the basis of a space.

The stop word filtering unit 220 generates stop word filtering data by filtering the stop word based on the natural language token. The stop word filtering unit 220 generates stop word filtering data by performing stop word filtering for selecting and removing the natural language tokens determined as predetermined stop words in the natural language token. Here, a stop word refers to a set of words that are predefined in order to remove a part corresponding to a token which is unnecessary in the analysis of a sentence or an expression. That is, in [Example 1] 'the' (other than a or to) is predefined as a dictionary in the system. Here, the dictionary means a list including a set of words. That is, after generating the natural language token, the natural language processing unit 130 performs a process of removing a stop word, which is a part unnecessary in the analysis. The stop word filtering is performed in a case where a mathematical problem is prolonged It prevents many tokens from entering and also works to improve the processing speed of the system. That is, the natural language and mathematical structuring apparatus 100 separates the first information constituting the natural language after the tokenizing process is performed using the stop word filtering unit 220 into a plurality of tokens, (100), the process of removing the stop word is performed in the next step. This eliminates unnecessary tokens to extract the semantic meaning. For example, 'i', 'low', 'here' and 'there' may be set as stop words, but not necessarily limited thereto. .

The de-duplication filtering unit 230 generates de-duplication filtering data in which de-duplication filtering is performed on the stop word filtering data. The de-duplication filtering unit 230 generates de-duplication filtering data by performing de-duplication filtering for selectively removing duplicated data from the stop word filtering data. That is, the natural language and mathematical formula structuring apparatus 100 performs a process of filtering a stop word using a de-duplication filtering unit 230 and removing redundancy, and by removing redundant words through de-duplication filtering, The processing load of the structuring apparatus 100 can be reduced.

The operation matching unit 240 matches operation information given a predetermined meaning to the deduplication filtering data. The operation matching unit 240 stores the data corresponding to the predicate in the deduplication filtering data and stores the matching operation information with the predefined meaning. Here, the operation information means summary information that can be extracted based on a natural language token or an expression token. For example, in [Example 1], action information called 'Solve' can be extracted based on a natural language token or a mathematical token. Here, the reason why the data corresponding to the predicate in the deduplication filtering data is matched with the operation information is that in the process of defining the combination data (mathematical problem) as a schema, information on the representative operation of the whole sentence is obtained This is to be used later as a tool to assist in analyzing the similarity between searches or problems. The natural language and mathematical structuring apparatus 100 analyzes the characteristics of the combination data through a pre-processing using the operation matching unit 240, compares the predefined meaning with the token, . That is, the natural language and mathematical formula structuring apparatus 100 may classify the mathematical expressions included in the combination data into 'condition', 'definition', or the like based on the result obtained by the natural language processing unit 130 using the operation matching unit 240 , And can be used to grasp the semantic meaning of the mathematical content itself.

3 is a block diagram schematically showing a mathematical expression processing unit according to an embodiment of the present invention.

The mathematical expression processing unit 140 according to an embodiment of the present invention includes a tree transform unit 310, a semantic parser 320, and a mathematical tokenizing unit 330. In the embodiment of the present invention, the mathematical expression processing unit 140 includes only the tree transform unit 310, the semantic parser unit 320, and the mathematical tokenizing unit 330. However, It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the essential characteristics of one embodiment of the present invention, Various modifications and variations may be applied to the elements. Here, semantics means to understand the meaning of specific information in a corresponding device and make logical reasoning possible.

The natural language and mathematical structuring apparatus 100 receives the individual expressions created in a standardized format through the information input unit 110 and transmits the individual expressions to the expression processing unit 140. That is, the formula transferred to the mathematical expression processing unit 140 is in the form of an XML tag based on Math ML (Mathematical Markup Language), which is a standard defined by the World Wide Web Consortium (W3C). However, the formula transferred to the mathematical expression processing unit 140 is preferably Math ML, but is not limited thereto.

The tree transform unit 310 transforms the equation into a tree form. The tree transforming unit 310 transforms the mathematical expression created in the Math ML into an XML tree form, and then converts it into a DOM form. The natural language and mathematical structuring apparatus 100 converts the mathematical expression into a MathML XML tree using the tree transform unit 310, and the tree is converted into a DOM and converted into a tree form accessible by the program.

The semantic parser 320 performs the transcoding process on the expression converted into the tree form. The semantic parser 320 performs the traversal as a depth-first search method in which the second information constituting the equation is gradually transferred from the lowermost node to the upper node. The natural language and mathematical structuring apparatus 100 traverses the semantic parser 320 in order to grasp the semantic meaning of the mathematical expression using the semantic parser 320. The semantic parser 320 sequentially transmits information from the lowest node to the upper node The traversal is performed with a depth-first search, which is a form of forwarding. As a result, all the second information collected through the semantic parser 320 is gathered at the highest node, and a token of a formula is formed based on this information. To describe the transversal process and depth-first search in detail, the equation is generally in the form of a Math ML, which is organized in the form of a tree. This process of traversing the tree is called a traversal process. When doing this, use Depth-First Search. The traversal process starts from the root of the tree and goes to the child node first. Then, when all the child nodes are searched, it moves to the parent node. Therefore, all the information held by the child node is transferred to the parent node. In terms of time complexity, it is efficient to search only the number of edges.

The mathematical tokenizing unit 330 generates a mathematical token in which tokenization is performed on the mathematical expression in which the traversal process is performed. Here, the mathematical token refers to individual unit information obtained after parsing the mathematical expression included in the combination data (mathematical problem). That is, a tokenized formula token is a token made up of a Mathematics Natural Language. On the other hand, formula tokens are treated differently than natural tokens. In other words, the natural language processing unit 130 matches the action based on the natural language token, whereas the mathematical expression token becomes an output result in the mathematical expression processing unit 140, and the mathematical token can be used for an operation have.

4 is a flowchart illustrating a method of structuring a natural language and an expression according to an embodiment of the present invention.

The natural language and mathematical structuring apparatus 100 receives the combination data composed of a combination of a natural language and a mathematical expression (S410). Here, the combination data composed of a combination of natural language and mathematical expression can be directly input by a user's operation or command, but not always limited thereto, and document data composed of natural language and combination of mathematical expressions may be inputted from a separate external server will be. The natural language and mathematical structuring apparatus 100 separates the natural language and the mathematical expression from the combined data, respectively (S420). That is, the natural language and mathematical structuring apparatus 100 separates and recognizes the natural language and the expression included in the combination data when the combination data composed of a combination of natural language and mathematical expression is input.

The natural language and mathematical formula structuring apparatus 100 analyzes the first information constituting the separated natural language and processes the information according to a specific meaning (S430). That is, the natural language and mathematical formula structuring apparatus 100 generates a natural language token obtained by tokenizing a natural language, generates word filtering data by filtering a stop word based on a natural language token, performs deduplication filtering on the stop word filtering data Generates the deduplication filtering data, and matches the predefined motion information to the deduplication filtering data. The natural language and mathematical structuring apparatus 100 generates a natural language token by performing tokenization on first information constituting a natural language. The natural language and mathematical formula structuring apparatus 100 generates stop word filtering data by performing stop word filtering for selecting and removing the natural language tokens identified as predetermined stop words in the natural language token. The natural language and mathematical structuring apparatus 100 performs deduplication filtering to selectively remove redundant data from the stop word filtering data to generate deduplication filtering data. The natural language and mathematical structuring apparatus 100 stores data corresponding to a predicate in the deduplication filtering data and stores it with operation information having a predefined meaning.

In step S440, the natural language and mathematical structuring apparatus 100 analyzes the second information constituting the separated expressions and classifies them according to a specific meaning. The natural language and mathematical structuring apparatus 100 converts a formula into a tree form, performs a transverse process on a formula transformed into a tree form, and performs tokenization on a formula in which a transit process is performed. The natural language and mathematical structuring apparatus 100 converts an expression written in Math ML into an XML tree form and then converts the expression into a DOM form. The natural language and mathematical structuring apparatus 100 executes the traversal as a depth-first search method in which the second information constituting the equation is gradually transferred from the lowermost node to the upper node.

The natural language and mathematical structuring apparatus 100 recombines at least one of the first information, the second information, the natural language, and the mathematical expression and stores it as the recombination data (S450). The natural language and mathematical structuring apparatus 100 converts the recombined data into document data. That is, by performing steps S410 to S450, natural language and mathematical expressions can be stored and managed as recombined data through the natural language and mathematical formula structuring apparatus 100, and the mathematical expression can be retrieved using the stored recombination data We can extract the semantics that follow.

Although it is described in FIG. 4 that steps S410 to S450 are sequentially executed, it is only an exemplary description of the technical idea of an embodiment of the present invention. It is to be understood that the technical knowledge in the technical field to which an embodiment of the present invention belongs Those skilled in the art will appreciate that various modifications and adaptations may be made to those skilled in the art without departing from the essential characteristics of one embodiment of the present invention by changing the order described in FIG. 4 or by executing one or more of steps S410 through S450 in parallel And therefore, it is not limited to the time-series order in Fig.

As described above, the method for structuring natural language and mathematical expressions according to an embodiment of the present invention shown in FIG. 4 can be implemented by a program and recorded on a computer readable recording medium. A program for implementing the natural language and expression structuring method according to an embodiment of the present invention is recorded, and a computer-readable recording medium includes all kinds of recording devices for storing data that can be read by a computer system. Examples of such computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, etc., and also implemented in the form of a carrier wave (e.g., transmission over the Internet) . The computer readable recording medium may also be distributed over a networked computer system so that computer readable code is stored and executed in a distributed manner. In addition, functional programs, code, and code segments for implementing an embodiment of the present invention may be easily inferred by programmers skilled in the art to which an embodiment of the present invention belongs.

5 is an exemplary diagram illustrating a tree-like representation of a formula according to an embodiment of the present invention.

As shown in FIG. 5, when a mathematical content has a structure, child nodes connected to a root node are classified into natural language and mathematical form while maintaining word order information, . In addition, each natural language has a special meaning according to the order of connection of sentences. In other words, most of the mathematical contents have structures that weave mathematical expressions based on natural language. For example, a formula that follows a natural word can be linked as a specific condition, or the structure of the following formula can be defined. The semantic meaning can be extracted by integrating not only the meaning and connection relation of the natural language of each node but also the natural language. In other words, in order to classify the motions such as whether the mathematical contents require solving the equations, or describing them, it can be used to grasp the meaning by grouping the whole natural words together to grasp the direction of the problem.

FIG. 6 is an exemplary diagram illustrating a system in which a natural language and expression structuring apparatus according to an embodiment of the present invention provides data to cloud computing.

A system including a terminal 610, a communication network 620, and a natural language and expression structuring apparatus 100 is required for providing natural language and mathematical structuring apparatuses according to an embodiment of the present invention with cloud computing.

Herein, the terminal 610 is a terminal capable of transmitting and receiving various data via the communication network 620 according to a command or an operation of a user. The terminal 610 may be a tablet PC, a laptop, a personal computer (PC) A personal computer, a smart phone, a personal digital assistant (PDA), and a mobile communication terminal. In addition, the terminal 610 may be a cloud computing terminal that supports cloud computing that can use services such as data reading, writing, storing, network, and content usage through the communication network 120. That is, the terminal 610 includes a memory for storing a program for accessing the natural language and mathematical formula structuring apparatus 100 via the communication network 620, a device having a microprocessor for executing and calculating a program, . That is, the terminal 610 may be any terminal as long as it is connected to the communication network 620 and is capable of server-client communication with the natural language and mathematical formula structuring apparatus 100, and may be any communication terminal such as a notebook computer, It is a broad concept that includes all of the devices. Meanwhile, the terminal 610 is preferably formed with a touch screen, but is not limited thereto.

The terminal 610 can organize natural language and mathematical expressions in a cloud computing manner through the natural language and mathematical expression structuring apparatus 100. [ That is, the terminal 610 is provided with a separate input / output interface (not shown) for providing an input / output interface with the storage medium stored in the natural language and expression structuring apparatus 100 for structuring the natural language and the expression from the natural language and mathematical formula structuring apparatus 100 to the cloud computing system. And an interface controller for reading and writing data to and from the storage medium stored in the natural language and mathematical formula structuring apparatus 100 through the input / output interface unit. More specifically, the terminal 610 can input, through the input / output interface unit, the combination data composed of a combination of natural language and mathematical expression to the natural language and mathematical expression structuring apparatus 100, And the first information constituting the separated natural language is analyzed and classified according to a specific meaning, and each second information constituting the separated expression is analyzed It is possible to generate / store the recombination data in which at least one of the first information, the second information, the natural language, and the formula is recombined, so that the terminal 610 can actually generate natural language and / It can be structured as an equation.

The communication network 620 is a network capable of transmitting and receiving data using an Internet protocol using various wired and wireless communication technologies such as an Internet network, an intranet network, a mobile communication network, and a satellite communication network. The network 620 includes a terminal 610, And performs the function of relaying the data between them. The communication network 620 is connected to the natural language and mathematical structuring apparatus 100 to store computing resources such as hardware and software and to provide a computing resource required by the client to the corresponding terminal 610. [ . ≪ / RTI >

The natural language and mathematical structuring apparatus 100 may allow the terminal 610 to provide data for the storage medium stored in the natural language and mathematical structuring apparatus 100 in order to enable natural language and mathematical expressions to be structured via the terminal 610 with cloud computing. When the combination data composed of a combination of natural language and mathematical expression is input, the natural language and the mathematical expression are separated from the combined data, and each first information constituting the separated natural language is analyzed to obtain a specific meaning , Separates the second information constituting the separated expression according to a specific meaning, generates recombination data in which at least one of the first information, the second information, the natural language, and the formula is recombined And transmits only a part of the data of the recording medium to the terminal 610, It can provide cloud computing that enables structured natural language and formulas without having to install the application from the end of 610. That is, the natural language and mathematical formula structuring apparatus 100 includes a storage unit for storing a storage medium for structuring natural language and mathematical expressions in a cloud computing manner, and a storage unit for storing the stored data in a cloud for allowing the terminal 610 to read and write data to the storage medium. A computing unit may be additionally provided.

7 is a diagram illustrating an example of a method of analyzing information constituting a natural language and a formula according to an embodiment of the present invention and classifying the information according to a specific meaning.

The natural language processing unit 130 and the mathematical expression processing unit 140 may be configured to perform operations of the natural language processing unit 130 and the mathematical expression processing unit 140 in order to grasp a specific meaning, After analyzing the information, a specific meaning can be grasped by using at least one or more of the structure of the sentence, the keyword included and the type of the expression, and the semantic information classified into the identified specific meaning can be generated.

The natural language processing unit 130 and the modification processing unit 140 operate based on a predetermined rule to grasp a specific meaning. Specifically, as shown in FIG. 7A, a natural language and a combination of mathematical expressions 7B, the natural language processing unit 130 and the mathematical expression processing unit 140 are arranged in the order of the mathematical expressions P1, P2, P3, and P4, (Parsed) result of the first information constituting the natural language and the second information constituting the expression can be generated.

For example, in the case of P1, analyzing the first information constituting the natural language by the natural language processing unit 130 indicates that the formula name (Name) is "Find" and the type is a verb (VB) 140), the equation (Equation) is true and the polynomial is true (True). As shown in (C) of FIG. 7, Compared with the logical condition of the stored rule, it can be seen that R1 matches R1 among R1, R2 and R3. Therefore, as shown in FIG. 7 (D), "Solve" which is operation information satisfying the logical condition can be extracted as the operation information from the matched rule. That is, in this case, the specific meaning indicated by P1 can be recognized and identified as the motion index.

The natural language processing unit 130 or the mathematical expression processing unit 140 can extract all the operation information satisfying the logical condition of the stored rule. If the logical condition of the combination of the natural language token and the mathematical token may satisfy various logical conditions of the stored rule, in this case, one mathematical problem includes a plurality of pieces of operation information, If the token combination does not satisfy any logical condition, the compound sentence can be judged to be omitted in the analysis of the mathematical sentence (combination data) when the rule is generated, or an item not included in the analysis process or an incorrect mathematical sentence. In addition, the natural language processing unit 130 or the mathematical expression processing unit 140 may match mathematical expressions that are the subject of the natural language tokens generated as a result of the natural language parsing, among mathematical tokens.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed according to the following claims, and all technical ideas falling within the scope of the same shall be construed as falling within the scope of the present invention.

As described above, the present invention is applied to various fields that enable management of natural language and mathematical expression data by recombining natural language and mathematical expression based on analysis contents in which natural language processing and mathematical expression processing are performed on combined data of natural language and mathematical expression, It is a useful invention that generates the effect that can be used as a keyword when searching for a mathematical content by using a standard document in which natural language and formula are combined together.

110: information input unit 120:
130: natural language processing unit 140:
150: Data management unit 210: Natural language tokenizing unit
220: Stop word filtering unit 230: Duplicate elimination filtering unit
240: Operation matching unit 310: Tree conversion unit
320: Semantic parser 330: Expression tokenizer
610: Terminal 620:

Claims (17)

An information input unit for receiving combination data composed of a combination of a natural language and a mathematical formula;
A separator for separating the natural language and the expression from the combination data;
A natural language processing unit for analyzing the natural language and generating natural language processing information classified according to a specific meaning;
A mathematical expression processing unit for analyzing the mathematical expression and generating mathematical expression processing information classified according to a specific meaning; And
A data management unit for recombining the natural language processing information and some information included in the mathematical expression processing information and storing the recombined data as recombination data;
Wherein the natural language and mathematical structuring apparatus comprises: The method according to claim 1,
The natural language processing unit,
A natural language tokenizing unit for generating a natural language token obtained by tokenizing the natural language;
A stop word filtering unit for generating stop word filtering data in which a stop word is filtered based on the natural language token;
A de-duplication filtering unit for generating de-duplication filtering data in which de-duplication filtering is performed on the stop word filtering data; And
An operation matching unit operable to match the operation information to which the meaning defined in the pre-
Wherein the natural language and mathematical structuring apparatus comprises: 3. The method of claim 2,
The natural language tokenizing unit,
And generates the natural language token by performing tokenization on the natural language. 3. The method of claim 2,
Wherein the stop word filtering unit comprises:
Wherein the stop word filtering unit generates stop word filtering data by performing stop word filtering for selecting and removing a natural language token identified as a predetermined stop word in the natural language token. 3. The method of claim 2,
Wherein the de-
Wherein the de-duplication filtering data is generated by performing de-duplication filtering for selectively removing duplicated data from the stop word filtering data. 3. The method of claim 2,
Wherein the operation-
Wherein the data corresponding to the predicate in the de-duplication filtering data is matched with the operation information having the predefined meaning. The method according to claim 1,
The above-
A tree transforming unit for transforming the equation into a tree form;
A semantic parser for performing a traverse process on the expression transformed into the tree form; And
A formula token generating unit for generating a mathematical token in which tokenization is performed on the mathematical expression in which the traversal process is performed,
Wherein the natural language and mathematical structuring apparatus comprises: 8. The method of claim 7,
Wherein the tree conversion unit comprises:
And converting the mathematical expression created in Math ML (Mathematical Markup Language) into an XML tree form and then converting the mathematical form into a DOM (Document Object Model). 8. The method of claim 7,
Wherein the semantic parser comprises:
Wherein the traversal is performed by a depth-first search method in which the expression is gradually transferred from an uppermost node to an uppermost node. The method according to claim 1,
The data management unit,
And converting the recombined data into document data. An information input unit for inputting combination data composed of a combination of a natural language and a formula;
A separator for separating the natural language and the expression from the combination data;
Generating stop word filtering data in which a stop word is filtered based on the natural language token, generating duplicate removal filtering data in which duplicate removal filtering is performed on the stop word filtering data, A natural language processing unit for generating natural language processing information by matching the operation information given with the predefined meaning to the duplicate removal filtering data;
A mathematical expression processing unit for analyzing the mathematical expression and generating mathematical expression processing information classified according to a specific meaning; And
A data management unit for recombining the natural language processing information and some information included in the mathematical expression processing information and storing the recombined data as recombination data;
Wherein the natural language and mathematical structuring apparatus comprises: Separating the natural language and the mathematical expression from the combined data when the combined data composed of a combination of natural language and mathematical expression is input, generating natural language processing information classified according to a specific meaning by analyzing the natural language, A storage unit for storing a program for generating mathematical expression processing information classified according to a specific meaning and generating recombination data in which the natural language processing information and some information included in the mathematical expression processing information are recombined; And
And a terminal for allowing the terminal to read and write data on the storage medium,
Wherein the natural language and mathematical structuring apparatus comprises: An information input unit for inputting combination data composed of a combination of a natural language and a formula;
A separator for separating the natural language and the expression from the combination data;
A natural language processing unit for analyzing the natural language and generating natural language processing information classified according to a specific meaning;
A mathematical expression processing unit for converting the mathematical expression into a tree form, performing a traverse process on the mathematical expression converted into the tree form, and generating a mathematical token in which tokenization is performed on the mathematical expression in which the traversal process is performed; And
A data management unit for recombining the natural language processing information and some information included in the mathematical expression token and storing the recombined data as recombination data;
Wherein the natural language and mathematical structuring apparatus comprises: An information input step of inputting combination data composed of a combination of a natural language and a formula;
A separating step of separating the natural language and the expression from the combination data, respectively;
A natural language processing step of analyzing the natural language and generating natural language processing information classified according to a specific meaning;
A mathematical expression processing step of analyzing the mathematical expression and generating mathematical expression processing information classified according to a specific meaning; And
A data management step of recombining the natural language processing information and some information included in the mathematical expression processing information and storing the recombined data as recombination data
Wherein the natural language and the mathematical expression are structured as follows. 15. The method of claim 14,
In the natural language processing step,
A natural language tokenization step of generating a natural language token by tokenizing the natural language;
A stop word filtering step of generating stop word filtering data in which a stop word is filtered based on the natural language token;
A de-duplication filtering step of generating de-duplication filtering data in which de-duplication filtering is performed on the stop word filtering data; And
An operation matching step of matching the operation information imparted with the predefined meaning to the de-duplication filtering data
Wherein the natural language and the mathematical expression are structured as follows. 15. The method of claim 14,
The above-
A tree transforming step of transforming the equation into a tree form;
A semantic parser step of performing a transitive process on the expression transformed into the tree form; And
A formula tokenization step of performing tokenization on the formula in which the transit process is performed
Wherein the natural language and the mathematical expression are structured as follows. A computer-readable recording medium having recorded thereon a program for executing each step of a natural language and a formula structuring method according to any one of claims 14 to 16.

Images