KR20200114555A - Apparatus and method for matching natural language based on morphological analysis and system for controlling electronic document using the same - Google Patents

Abstract

Disclosed are a device and method for matching a natural language based on a morphological analysis and a system for controlling an electronic document using the same. According to one embodiment of the present invention, the device for matching a natural language based on a morphological analysis may comprise: a pattern generation unit acquiring concepts included in a plurality of sentences through a morpheme analysis of a plurality of sentences containing similar intentions to create a pattern string and generating a pattern based on the pattern string; and a user input analysis unit receiving a user input and matching the pattern corresponding to the user input to return matching data.

Description

A natural language matching device and method based on morpheme analysis, and an electronic document control system using the same {APPARATUS AND METHOD FOR MATCHING NATURAL LANGUAGE BASED ON MORPHOLOGICAL ANALYSIS AND SYSTEM FOR CONTROLLING ELECTRONIC DOCUMENT USING THE SAME}

The present application relates to a natural language matching apparatus and method based on morpheme analysis, and an electronic document control system using the same.

In general, various forms of documents are generated in the business area, and if these documents were prepared in the form of printed on paper in the past, electronic documents and electronic signature services that generate various documents in electronic form have been recently introduced and utilized. , It is showing a rapid growth trend in line with the government's paperless policy.

In addition, with the popularization of smart devices, user experiences using voice-based services have been continuously accumulated. Currently, most of the voice-based services provide functions related to the daily life of individuals, and are expected to expand to the business area of companies in the future.

Accordingly, there is an increasing need for a technology and a system for generating an electronic document based on various types of input including a voice signal, or a technology and a system capable of controlling and managing an electronic document based on a voice input.

In addition, in order to control and manage electronic documents based on various types of input including voice signals, the user's intention must be identified by analyzing the input, and in the case of input made in natural language (for example, voice input, text input) In addition, rule-based matching may be used to determine whether a specific word or phrase exists or is repeated in a natural language.

For the above-described rule-based matching, scripts and regular expressions are used in most development languages. In the case of Hangul, a total of 11172 syllables can be expressed. Since the number of syllables is so large, even if only one of the elementary/medium/final constituents of the syllable is changed, a greatly changed syllable is derived, so it is a difficult problem to describe a regular expression so that similar syllables can all match.

In addition, in the case of Hangul, roots and affixes can be easily combined. In particular, various affixes can be used in the case of verbs and adjectives that are verbs. In order to match all conjugable conjugations for one basic type, a pattern that can cover even irregular phenomena in which specific characters/vowels are dropped or added when combined with an affix must be created.

The technology behind the present application is disclosed in Korean Patent Publication No. 10-1604553.

The present application is to solve the problems of the prior art described above, and an object thereof is to provide a method and apparatus capable of matching natural language by introducing morpheme analysis.

The present application provides a natural language matching method and apparatus capable of matching all natural languages including derived words by analyzing a plurality of sentences containing similar intentions, extracting roots, and reflecting them in a pattern as to solve the problems of the prior art described above. It is intended to be provided.

The present application is to solve the problems of the prior art described above, grasps the intention of the user who generated the user input based on the pattern, and performs control operations such as creation, processing, and search for electronic documents according to the user's intention. It aims to provide an electronic document control system based on natural language processing.

However, the technical problem to be achieved by the embodiments of the present application is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, the morpheme analysis-based natural language matching apparatus according to an embodiment of the present application includes a concept included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions. And a pattern generator configured to obtain a pattern string and generate a pattern based on the pattern string, and a user input analyzer configured to receive a user input and match the pattern corresponding to the user input to return matching data. have.

In addition, the pattern generation unit may include a morpheme analysis unit that performs morpheme analysis on each of the plurality of sentences, a concept acquisition unit that acquires a concept based on a result of the morpheme analysis, and a capture or the above for extracting a specific phrase from the user input. It may include a pattern character string creating unit for creating a pattern character string including at least one of the concepts and a pattern defining unit defining a pattern based on the pattern character string.

In addition, the morpheme analysis result may include a morpheme obtained by analyzing each of the plurality of sentences in a morpheme unit, and part of speech information associated with the analyzed morpheme.

In addition, the POS information may be displayed based on a preset POS tag.

In addition, the pattern string creation unit may create the pattern string based on a concept set including at least one concept obtained by the concept acquisition unit.

In addition, the concept may be defined in the form of an anonymous concept in which a concept name encompassing a word that is an element of the concept exists and is defined based on the concept name, or a word that is an element of the concept is directly listed without a separate concept name. .

In addition, the user input analysis unit may include a segmentation unit that divides the user input into a plurality of segments by cutting the user input into spaces, and determines whether the at least one segment matches the pattern, and if the matching is successful, matching data It may include a matching execution unit that returns.

In addition, the matching performing unit may perform matching of at least one segment and a word or matching of the at least one segment and a concept.

In addition, the matching unit may determine whether to perform matching of a word or concept described in a pattern string, and return matching data based on whether the matching is performed.

In addition, in order to match the concept with at least one segment, the matching unit determines whether a concept having a word matching one segment as an element exists, and if not, including a space, the one segment and the next It is possible to match concepts that combine segments and match words as elements.

In addition, the matching unit may sequentially combine segments to determine whether or not an element in the concept matches, and if there is no matching element, it may determine whether to match an element of the next concept.

In addition, when a capture is included in the pattern string, the matching unit may match the capture with the at least one segment to extract a specific phrase in the user input associated with the capture.

In addition, the matching data may include at least one of data on the matched concept or a specific phrase in a user input associated with the capture.

On the other hand, the electronic document control system based on natural language processing according to an embodiment of the present application receives a user input, matches a pattern corresponding to the user input, returns matching data, and is embedded in the user input based on the matching data. A natural language matching device that grasps the user's intention, an electronic document creation device that creates or corrects an electronic document based on an electronic format template according to the user's intention, and an electronic document that has been previously created according to the user's intention. An electronic document processing device that performs approval, reception, or deletion processing, an electronic document search device that provides a list of electronic documents that meet the search conditions included in the user input according to the user's intention, and a plurality of electronic devices according to the user's intention It may include a navigation module that controls movement between documents.

On the other hand, the natural language matching method based on morpheme analysis according to an embodiment of the present application is included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions, and based on the result of morpheme analysis, Acquiring a concept having a configured element, creating a pattern string, generating a pattern based on the pattern string, receiving a user input, and returning matching data by matching the pattern corresponding to the user input. Can include.

In addition, the generating of the pattern may include performing a morpheme analysis on each of the plurality of sentences, acquiring a concept based on a result of morpheme analysis, a capture for extracting a specific phrase from the user input, or among the concepts It may include creating a pattern string including at least one and defining a pattern based on the pattern string.

In addition, the returning of the matching data may include dividing the user input into a plurality of segments by cutting the received user input in units of spaces, and determining whether at least one segment matches the pattern and matching is successful. In this case, it may include generating matching data.

In addition, the generating the matching data may include matching at least one segment and a word or matching the at least one segment and a concept, determining a pattern string including all the matched words and the matched concept, When the pattern string includes a capture, extracting a specific phrase in a user input associated with the capture by matching the capture and the at least one segment, and data on the matched concept or a user input associated with the capture It may include generating the matching data including at least one of my specific phrases.

The above-described problem solving means are merely exemplary and should not be construed as limiting the present application. In addition to the above-described exemplary embodiments, additional embodiments may exist in the drawings and detailed description of the invention.

According to the above-described problem solving means of the present application, it is possible to provide a method and apparatus capable of matching natural language by introducing morpheme analysis.

According to the above-described problem solving means of the present application, it is possible to provide a natural language matching method and apparatus that can match all natural languages including derived words by analyzing a plurality of sentences containing similar intentions, extracting the root, and reflecting it in the pattern. It works.

According to the above-described problem solving means of the present application, a natural language capable of grasping the intention of the user who generated the user input based on the pattern, and performing control operations such as creation, processing, and search for an electronic document according to the user's intention. There is an effect of providing a processing-based electronic document control system.

However, the effect obtainable in the present application is not limited to the effects as described above, and other effects may exist.

1 is a diagram illustrating a configuration of an electronic document control system based on natural language processing according to an embodiment of the present application.
2 is a diagram illustrating a configuration of a natural language matching device based on morpheme analysis according to an embodiment of the present application.
3 is a diagram illustrating an example of a preset POS tag classified according to detailed parts of speech according to a Hangul morpheme according to an embodiment of the present application.
4 is a diagram showing a result of morpheme analysis for a plurality of sentences containing similar intentions according to an embodiment of the present application.
5 is a diagram illustrating a result of detecting a morpheme corresponding to a root root from a result of morpheme analysis according to an embodiment of the present application.
6 is a diagram showing a result of classifying a morpheme corresponding to a root according to an embodiment of the present application into a body language and a verb.
7 is a diagram illustrating a method of acquiring a concept according to an embodiment of the present application.
8 is a diagram for describing a pattern character string according to an embodiment of the present application.
9 is a diagram for explaining a pattern based on a pattern character string according to an embodiment of the present application.
10 is a diagram showing the configuration of a pattern generator according to an embodiment of the present application.
11 is a diagram showing the configuration of a user input analysis unit according to an embodiment of the present application.
12 is an operation flow diagram of a natural language matching method based on morpheme analysis according to an embodiment of the present application.
13 is an operation flow diagram of a method for generating a pattern according to an embodiment of the present application.
14 is a flowchart illustrating an operation of a method for generating matching data according to an embodiment of the present application.

Hereinafter, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present application. However, the present application may be implemented in various different forms and is not limited to the embodiments described herein. In addition, in the drawings, parts not related to the description are omitted in order to clearly describe the present application, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the present specification, when a part is said to be "connected" with another part, it is not only "directly connected", but also "electrically connected" or "indirectly connected" with another element interposed therebetween. "Including the case.

Throughout this specification, when a member is positioned "on", "upper", "upper", "under", "lower", and "lower" of another member, this means that a member is located on another member. It includes not only the case where they are in contact but also the case where another member exists between the two members.

Throughout the specification of the present application, when a certain part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

1 is a diagram showing the configuration of an electronic document control system based on natural language processing according to an embodiment of the present application.

Referring to FIG. 1, a natural language processing-based electronic document control system 10 according to an embodiment of the present application includes a morpheme analysis-based natural language matching device 100, an electronic document creation device 200, and an electronic document processing device 300. , An electronic document search device 400 and a navigation module 500 may be included.

The morpheme analysis-based natural language matching apparatus 100 according to an embodiment of the present application receives a user input, matches a pattern corresponding to the user input, returns matching data, and includes the matching data in the user input. You can understand the intention of the user.

In addition, according to the exemplary embodiment of the present application, the natural language matching apparatus 100 based on morpheme analysis may receive the user input in the form of a voice input in which a sentence spoken by the user as the user input is used as the user input.

In addition, according to another embodiment of the present application, the natural language matching device 100 based on morpheme analysis may input the user input in the form of an interactive input in which the user inputs a sentence typed through a separate input device such as a keyboard. Can receive.

In addition, according to an embodiment of the present application, the user's intention includes the creation or creation of a new electronic document such as an intention to create a document, an intention to modify a document, an intention to approve a document, an intention to receive a document, an intention to delete a document, an intention to search a document, or an intention to move a document. It may include orders or requests for overall business procedures using electronic documents, such as management of written electronic documents.

The electronic document creation apparatus 200 may create or modify an electronic document based on the electronic form template according to the intention of the user.

According to an embodiment of the present application, when the user's intention identified by the morpheme analysis-based natural language matching device 100 corresponds to the document writing intention, the electronic document writing device 200 corresponds to the document writing intention. The electronic form template may be loaded, and data extracted from a user input in units of a key value and a value value may be input into the entry item in the electronic form template.

The electronic document processing apparatus 300 may perform approval, reception, or deletion processing for the previously created electronic document 2 according to the intention of the user.

In addition, according to an exemplary embodiment of the present disclosure, the electronic document processing apparatus 300 may transmit an electronic document approval request through a push notification or an alarm, and may provide a list of electronic documents associated with the approval request.

The electronic document search apparatus 400 may provide a list of electronic documents meeting a search condition included in a user input according to the intention of the user.

According to the exemplary embodiment of the present disclosure, the electronic document search apparatus 400 may obtain a search condition based on the user input.

According to an embodiment of the present application, the electronic document search apparatus 400 collects electronic documents meeting the search conditions from an electronic document database in which a previously created electronic document 2 is stored and creates a list, and the electronic document list Can provide.

The navigation module 500, Movement between a plurality of electronic documents may be controlled according to the user's intention.

The natural language matching device 100 and the electronic document creation device 200, the electronic document processing device 300, the electronic document search device 400, and the navigation module 500 may communicate through a network 1 that is interconnected.

The network 1 is a wired and wireless network capable of exchanging information between the natural language matching device 100 and the electronic document creation device 200, the electronic document processing device 300, the electronic document search device 400, and the navigation module 500. It refers to a connection structure, and illustratively, 3GPP (3rd Generation Partnership Project) network, LTE (Long Term Evolution) network, 5G network, WIMAX (World Interoperability for Microwave Access) network, Internet, LAN (Local Area Network) ), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), Bluetooth (Bluetooth) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc. It is not limited thereto.

2 is a diagram illustrating a configuration of a natural language matching device based on morpheme analysis according to an embodiment of the present application.

Referring to FIG. 2, a morpheme analysis-based natural language matching apparatus 100 according to an exemplary embodiment of the present disclosure may include a pattern generation unit 110 and a user input analysis unit 120.

The pattern generator 110 may generate a pattern string by obtaining a concept included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions, and generate a pattern based on the pattern string.

In addition, according to the exemplary embodiment of the present application, the pattern generator 110 may perform morpheme analysis for each of the plurality of sentences.

In addition, according to an exemplary embodiment of the present disclosure, the pattern generator 110 may acquire a concept based on a result of morpheme analysis.

According to the exemplary embodiment of the present disclosure, the concept may be understood as a collection of minimum prior information that is referred to in the process of matching the user input with the pattern to be described later by the user input analysis unit 120. Specifically, the concept may be a target for which it is determined whether or not matching with one or more segments constituting a user input in a process of matching a pattern to be described later.

In addition, according to an embodiment of the present application, the morpheme analysis result may include a morpheme obtained by analyzing each of the plurality of sentences in a morpheme unit and part of speech information associated with the analyzed morpheme, and the part of speech information is a preset part of speech. It can be marked based on tags.

3 is a diagram illustrating an example of a preset POS tag classified according to detailed parts of speech according to a Hangul morpheme according to an embodiment of the present application.

Referring to FIG. 3, a preset POS tag 32 according to an embodiment of the present application includes parts of speech in Hangul morphemes, such as body language, verb, tube sentence, adverb, interjection, investigation, ending, affix, root, sign, Korean, etc. It may be divided into a large classification 31, classified according to detailed parts of speech 33 such as general nouns, proper nouns, dependent nouns, rhetoric, etc., and marked in uppercase alphabets according to a predetermined rule.

However, it should be understood that the preset POS tag shown in FIG. 3 is exemplary, and may be determined differently according to a morpheme analyzer used for morpheme analysis.

4 is a diagram showing a result of morpheme analysis for a plurality of sentences containing similar intentions according to an embodiment of the present application.

Referring to FIG. 4, a plurality of sentences 41 containing similar intentions may contain a'document creation intention', and the pattern generator 110 analyzes each of the plurality of sentences 41 in units of morphemes. , Part of speech information associated with the analyzed morpheme may be marked 42 together with the analyzed morpheme based on a preset POS tag.

According to an embodiment of the present application, the pattern generation unit 110 may divide a plurality of sentences into spaces (/), and provide the morpheme analysis result in the form of (analyzed morpheme):(part of speech tag). The morpheme of can be connected through','.

In addition, according to an exemplary embodiment of the present application, the pattern generator 110 may detect a morpheme corresponding to a root root from the result of the morpheme analysis, and classify the morpheme corresponding to the root root into a body language and a verb.

5 is a diagram illustrating a result of detecting a morpheme corresponding to a root root from a result of morpheme analysis according to an embodiment of the present application.

Referring to FIG. 5, the morpheme 50 corresponding to the root may mean a morpheme displayed in bold. According to an embodiment of the present application, the morpheme 50 corresponding to the root root is a noun series (NNx form based on a part of speech tag) ), adjective series (VA), or verb series (VV).

According to an exemplary embodiment of the present disclosure, morphemes (parts not shown in bold in FIG. 5) excluding the morpheme 50 corresponding to the root of the morpheme analysis result may not be utilized in the process of obtaining a concept to be described later.

6 is a diagram showing a result of classifying a morpheme corresponding to a root according to an embodiment of the present application into a body language and a verb.

Referring to FIG. 6, the morpheme 50 corresponding to the root can be classified into a body language side 61 and a word language side 62. According to an embodiment of the present application, the body word 61 may include a morpheme 50 corresponding to a root belonging to a noun, a pronoun, and a rhetoric first, and the verb side 62 corresponds to an adjective, a root belonging to a verb The morpheme 50 may be included preferentially, but even if it corresponds to a body language (ie, a noun, a pronoun, or a rhetoric), a portion having a high predicate tendency within a sentence may be classified as a proverb 62.

According to an embodiment of the present application, one or more morphemes classified as body language side or proverb side in one sentence may be defined as a word 60 encompassing the one or more morphemes, and the word 60 is an element of the concept. Can be.

According to an embodiment of the present application, the word 60 is a string word in which the morpheme is displayed as it is, or a tag word in which the analyzed morpheme is displayed together with the analyzed morpheme based on a preset part-of-speech tag. Can contain

In addition, according to an exemplary embodiment of the present disclosure, in the case of the character string word, in the process of matching a user input and a pattern to be described later, it may be regarded as successful only when it exactly matches a segment of the user input.

In addition, according to an embodiment of the present application, in the case of the tag word, when the morpheme and part-of-speech information described in the tag word exist in the result of analyzing the segment of the user input in the process of matching a user input and a pattern to be described later. , It may be considered that the segment of the user input and the tag word have been successfully matched.

According to an embodiment of the present application, the word 60 is a plurality of morphemes 50 of'write:NNG','low:VV', and'want:VX' like the word on the proverb for the third sentence of FIG. ) May be included.

In addition, according to an exemplary embodiment of the present application, the pattern generation unit 110 generates a concept of a word on the side of the body including the morphemes classified as the word, and a word on the side of the word including the morphemes classified as the word. It is possible to acquire the concept of the word of the word as an element.

For example, referring to FIG. 6, the concept of the word'document:NNG' and the word'document:NNP' may be used as elements, and the concept of the word'write:NNG' and the word'write: VV, G: NNB', the word'Write: NNG, Ha: VV,: Want: VX', the word 'Write: VV, Ha: VX', the word' Write: VV, Want: VX' as elements. have.

According to an embodiment of the present application, in the case of a body language, the concept of the body language side, which has a word on the body language as an element, may be set to preferentially match with one segment of the user input, due to the characteristics of a Hangul sentence structure with a little change in type. In the case of, the proverb-side concept using the proverb-side word as an element may be set to preferentially match two or more segments of the user input due to the characteristics of the Hangul sentence structure with various types of change.

In addition, according to an embodiment of the present application, the concept includes a concept name including a word that is an element of the concept, and is defined based on the concept name or directly listing the word as an element of the concept without a separate concept name. It can be defined in the form of an anonymous concept.

7 is a diagram illustrating a method of acquiring a concept according to an embodiment of the present application.

Referring to FIG. 7, the concept 71a and 71b in which the concept name exists has a concept name (for example, a document or written) encompassing a word as an element, and may be defined based on the concept name, and is anonymous. Concepts (72a, 72b) defined in concept form can be defined in a way that directly lists the elements of the concept (e.g., document:NN, creation:NN, write:VV, geo:NNB, etc.) without a separate concept name. I can.

In addition, as described above, the concept may be a word (a group of one or more morphemes) as an element, or another concept may be used as an element.

For example, the {~document} concept may include a word such as'document:NNP' as an element, and it is a sub-concept and includes pre-generated concepts such as {~report}, {~goodwill}, and {~request}. It can also be included as an element.

According to an embodiment of the present application, the element of the concept may be at least one of the string word, the tag word, or another concept, and if the concept is a different concept as an element, whether the concept name of each concept exists Need not be determined identically.

In addition, according to an embodiment of the present application, the pattern generation unit 110 generates a pattern string including at least one of the concept or a capture substituted to indicate the position of the specific phrase in order to extract the specific phrase from the user input. You can write it.

According to an embodiment of the present application, the pattern string is mainly written for the same purpose as a regular expression (regex), which is a kind of formal language used in a programming field related to a string, and the pattern generator 110 is obtained. It may include at least one of the concept or the capture.

In addition, the capture is {_*} in the pattern string to indicate the location of the specific phrase in order to extract the specific phrase from the user input in the process of matching the user input and the pattern by the user input analysis unit 120 It may mean a part that is replaced with a shape. However, the capture indicated by {_*} throughout the present application is not limited to the notation of the {_*} form, and may have various forms.

8 is a diagram for describing a pattern character string according to an embodiment of the present application.

Referring to FIG. 8, the pattern strings 80a and 80b according to the exemplary embodiment of the present disclosure may include at least one of a capture 81 and a concept 82.

According to an embodiment of the present application, the capture of the pattern string 80a shown on the upper side is a concept included in the pattern string such as a'planning' document, a'proof' document, and a'vacation application' document ({~[document: NN}). The character string itself (for example, planning, proof, and vacation request), which is placed in front of} and functions to limit the type of document, may be replaced with {_*} so that the user input can be obtained.

According to another embodiment of the present application, the capture of the pattern string 80b shown at the bottom is a concept encompassing the quantity units included in the pattern string such as'one','two' trees, and'three' times {~unit} It may be replaced with a {_*} format so that the quantity information preceding in can be obtained from a user input.

According to an embodiment, when there is no part to extract a specific phrase in a sentence, the capture may not be included in the pattern string.

In addition, according to the exemplary embodiment of the present application, the pattern generation unit 110 may obtain a concept described in the pattern string based on a concept set including at least one concept from the pattern string. In this case, the acquired concept may be used as a criterion for determining whether matching is successful in a process of matching a user input and a pattern to be described later.

In addition, according to the exemplary embodiment of the present application, the pattern generator 110 may create the pattern string based on a concept set including at least one concept.

That is, the pattern generation unit 110 does not need to newly define the concept for the word side or the word side of a plurality of sentences each time in the process of creating the pattern string, and calls the concept existing in the concept set. Pattern strings can be created, simplifying the processing of operations required to create patterns.

In addition, according to an exemplary embodiment of the present disclosure, the pattern generator 110 may define a pattern based on the pattern string.

9 is a diagram for explaining a pattern based on a pattern character string according to an embodiment of the present application.

Referring to FIG. 9, the patterns 90a and 90b according to an embodiment of the present application include a pattern string 80, a name 91 of a pattern determined according to similar intentions contained in the plurality of sentences, and the pattern string. It may include a language type 92 used for writing and a concept set 93 to which a concept included in the pattern string belongs.

According to an embodiment of the present application, the language type 92 may be expressed in a locale, and indicates the type of language used to create the pattern string so that the pattern can be defined even in languages other than Korean. , According to a predetermined rule, the language type may be indicated (eg, ko_KR for Korean).

According to an exemplary embodiment of the present disclosure, the concept set 93 represents a concept set to which a concept included in the pattern string 80 belongs, and may include all concepts that can be utilized when defining a pattern.

According to the exemplary embodiment of the present disclosure, the concept set 93 may be understood as an input transmitted to find a concept specified in a pattern string and set it in a pattern in the process of generating a pattern.

According to the exemplary embodiment of the present application, the concept set 93 may indicate a concept group in which a plurality of concepts are grouped according to a predetermined criterion. For example, when defining a corresponding pattern, the concept set 93 may be specified to indicate whether a pattern string of a corresponding pattern is created based on a concept belonging to which concept group.

In addition, according to an embodiment of the present application, in the process of determining whether the concept specified in the pattern string matches a user input to be described later, when the concept specified in the pattern string exists in the concept set 93, the concept is a concept set. It can be fetched from (93), and used in the process of determining whether to match a user input.

The user input analysis unit 120 may receive a user input, match the pattern corresponding to the user input, and return matching data.

In addition, according to the exemplary embodiment of the present application, the user input analysis unit 120 may receive the user input in the form of a voice input in which a sentence spoken by the user as the user input is used as the user input.

According to an exemplary embodiment of the present disclosure, the user input analysis unit 120 may recognize a voice spoken by a user and convert it into a text format.

To this end, the user input analysis unit 120 may use at least one speech recognition algorithm widely known in the art, and a speech-to-text (STT) technique may be used for conversion into the text format. have.

Further, according to another embodiment of the present application, the user input analysis unit 120 may receive the user input in the form of an interactive input in which the user inputs a sentence typed through a separate input device such as a keyboard. have.

In addition, according to an exemplary embodiment of the present disclosure, the user input analysis unit 120 may divide the user input into a plurality of segments by cutting the user input into spaces.

In addition, according to an exemplary embodiment of the present disclosure, the user input analysis unit 120 may determine whether at least one segment matches the pattern and return matching data when the matching is successful.

Further, according to an embodiment of the present application, the user input analysis unit 120 matches at least one segment and a word, or matches the at least one segment and a concept, and includes both the matched word and the matched concept. Matching data can be returned by determining a string.

In addition, according to an embodiment of the present application, in order to match the concept with at least one segment, the user input analysis unit 120 determines whether a concept having a word matching one segment as an element exists, and if not A concept in which a word that matches the combination of the one segment and the next segment as an element may be matched. The user input analysis unit 120 may sequentially combine the segments to determine whether or not they match the elements in the concept, and if there is no matching element, it may determine whether or not the elements of the next concept are matched.

According to an exemplary embodiment of the present disclosure, when a concept specified in a pattern string exists in a concept set, the user input analysis unit 120 may take it and use it for matching.

In addition, according to an embodiment of the present application, when a capture is included in the pattern string, the user input analysis unit 120 matches the capture with the at least one segment to extract a specific phrase in the user input associated with the capture. can do.

In addition, according to an exemplary embodiment of the present disclosure, the matching data may include at least one of data on the matched concept or a specific phrase in a user input associated with the capture.

10 is a diagram showing the configuration of a pattern generator according to an embodiment of the present application.

Referring to FIG. 10, the pattern generation unit 110 according to an embodiment of the present application includes a morpheme analysis unit 111, a concept acquisition unit 112, a pattern string creation unit 113, and a pattern definition unit 114 can do.

The morpheme analysis unit 111 may perform morpheme analysis for each of the plurality of sentences.

In addition, according to an exemplary embodiment of the present application, the morpheme analysis unit 111 may detect a morpheme corresponding to a root from the result of the morpheme analysis, and classify the morpheme corresponding to the root into a body language and a verb.

The concept acquisition unit 112 may acquire a concept based on the morpheme analysis result.

According to an embodiment of the present application, one or more morphemes classified as body language side or proverb side in one sentence may be defined as a word 60 encompassing the one or more morphemes, and the word 60 is an element of the concept. Can be.

In addition, according to an embodiment of the present application, the concept acquisition unit 112 includes a word on the side of the word including the word on the side of the body, including the morphemes classified as the word, as an element It is possible to acquire the concept of the word of the word as an element.

The pattern string creation unit 113 may create a pattern string including at least one of the concept or capture substituted to indicate the location of the specific phrase in order to extract the specific phrase from the user input.

In addition, according to an exemplary embodiment of the present disclosure, the pattern string creation unit 113 may create the pattern string based on a concept set including at least one concept acquired by the concept acquisition unit 112.

The pattern definition unit 114 may define a pattern based on the pattern character string.

In addition, according to an embodiment of the present application, the pattern definition unit 114 includes a pattern character string, a name of a pattern determined according to similar intentions contained in a plurality of sentences, a language type used to create the pattern character string, and the pattern character string. It may be to define the pattern including a concept set to which the concept included in belongs.

11 is a diagram showing the configuration of a user input analysis unit according to an embodiment of the present application.

Referring to FIG. 11, the user input analysis unit 120 according to the exemplary embodiment of the present disclosure may include a segmentation unit 121 and a matching execution unit 122.

The segmentation unit 121 may divide the user input into a plurality of segments by cutting the user input into spaces.

The matching performing unit 122 may determine whether or not the at least one segment matches the pattern, and may return matching data when the matching is successful.

According to an embodiment of the present application, the matching execution unit 122 matches at least one segment and a word, or matches the at least one segment and a concept, and determines a pattern string including both the matched word and the matched concept. To return the matching data.

In addition, according to an exemplary embodiment of the present application, when the pattern string includes a capture, the matching execution unit 122 matches the capture and the at least one segment to extract a specific phrase in the user input associated with the capture. I can.

In order to match the concept with at least one segment, the matching execution unit 122 determines whether a concept having a word matching one segment as an element exists, and if not, combines the one segment and the next segment. It is possible to match concepts with words that match those as elements. In addition, the matching execution unit 122 may sequentially combine the segments to determine whether or not they match the elements in the concept, and if there is no matching element, it may determine whether to match the element of the next concept.

In addition, according to an exemplary embodiment of the present disclosure, the matching data may include at least one of data on the matched concept or a specific phrase in a user input associated with the capture.

According to an embodiment of the present application, when the user input is the user's utterance of'I am going to write a vacation application document', the segmentation unit 121 uses the user input as a ‘vacation application document/to write/year’ It can be cut, and the first segment'vacation application document', the second segment'to write', and the third segment'sun' can be derived. That is, each of the cut parts based on the spacing unit can be a segment. .

Subsequently, the matching execution unit 122 may match a concept having a word matching the first segment to the third segment as an element, and as a result, the first segment may be matched with the concept {~document}, and the second The segment and the third segment alone do not match the concept, but the combination of the second segment and the third segment may match the concept {~writing}.

Subsequently, the matching execution unit 122 may determine a pattern string including the {~document} and {~write} concepts, and when the capture {_*} is indicated in front of the {~document} concept in the pattern string, A specific phrase called'vacation request', which is data corresponding to the capture, can be extracted from the user input.

Accordingly, the matching data may include information indicating that there is a part matching the concept {~document} and the concept {~write} from the user input, and a specific phrase “request for vacation”.

Accordingly, the natural language matching device 100 based on morpheme analysis according to an embodiment of the present application is based on the matching data, the intention of the user contained in the user input (the point is that it is a speech with the intention to write a vacation application document) May be recognized, and according to the user's intention, the electronic document creation device 200 may start to create an electronic document based on the electronic form template by calling the vacation application document electronic form template.

12 is an operation flow diagram of a natural language matching method based on morpheme analysis according to an embodiment of the present application.

The morpheme analysis-based natural language matching method illustrated in FIG. 12 may be performed by the morpheme analysis-based natural language matching apparatus 100 described above. Therefore, even if omitted below, the description of the natural language matching apparatus 100 based on morpheme analysis may be equally applied to FIG. 12.

Referring to FIG. 12, in step S1210, the pattern generator 110 obtains a concept included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions, creates a pattern character string, and creates a pattern character string in the pattern character string. You can create a pattern based on it.

Next, in step S1220, the user input analysis unit 120 may receive a user input.

Next, in step S1230, the user input analyzer 120 may match the pattern corresponding to the user input and return matching data.

According to an embodiment of the present application, the returning the matching data shown in FIG. 12 (S1230) comprises dividing the user input into a plurality of segments by cutting the received user input by spaces, and at least one segment and the It may include determining whether to match the pattern and generating matching data when the matching is successful.

In the above description, steps S1210 to S1230 may be further divided into additional steps or may be combined into fewer steps, according to an embodiment of the present disclosure. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

13 is an operation flow diagram of a method for generating a pattern according to an embodiment of the present application.

The pattern generation method illustrated in FIG. 13 may be performed by the natural language matching apparatus 100 based on morpheme analysis described above. Therefore, even if omitted below, the description of the natural language matching apparatus 100 based on morpheme analysis may be equally applied to FIG. 13.

Referring to FIG. 13, in step S1310, the morpheme analysis unit 111 may perform morpheme analysis for each of the plurality of sentences.

Next, in step S1320, the consap acquisition unit 112 may acquire a concept based on the morpheme analysis result.

Next, in step S1330, the pattern string creation unit 113 may create a pattern string including at least one of the concept or capture for extracting a specific phrase from the user input.

Next, in step S1340, the pattern definition unit 114 may define a pattern based on the pattern character string.

In the above description, steps S1310 to S1340 may be further divided into additional steps or may be combined into fewer steps, according to an embodiment of the present disclosure. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

14 is a flowchart illustrating an operation of a method for generating matching data according to an embodiment of the present application.

The method of generating matching data illustrated in FIG. 14 may be performed by the natural language matching apparatus 100 based on morpheme analysis described above. Accordingly, even if omitted below, the description of the natural language matching apparatus 100 based on morpheme analysis may be equally applied to FIG. 14.

Referring to FIG. 14, in step S1410, the segmentation unit 121 may divide the received user input into a plurality of segments by cutting the received user input into spaces.

Next, in step S1420, the matching execution unit 122 may match at least one segment and a word or match the at least one segment and a concept.

Next, in step S1430, the matching execution unit 122 may determine a pattern string including both the matched word and the matched concept.

Next, in step S1440, the matching performing unit 122 may determine whether a capture is included in the pattern string.

Next, in step S1450, when a capture is included in the pattern string, the matching execution unit 122 may match the capture and the at least one segment to extract a specific phrase in the user input associated with the capture.

Conversely, when the pattern string does not contain capture in step S1440, the matching execution unit 122 may proceed to step S1460, which will be described later.

Next, in step S1460, the matching execution unit 122 may generate the matching data including at least one of data on the matched concept or a specific phrase in the user input associated with the capture.

In the above description, steps S1410 to S1460 may be further divided into additional steps or may be combined into fewer steps, according to an embodiment of the present disclosure. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

The natural language matching method based on morpheme analysis according to an embodiment of the present application may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The above-described hardware device may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

In addition, the above-described morpheme analysis-based natural language matching method may be implemented in the form of a computer program or application executed by a computer stored in a recording medium.

The foregoing description of the present application is for illustrative purposes only, and those of ordinary skill in the art to which the present application pertains will be able to understand that it is possible to easily transform it into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present application is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present application.

10: Electronic document control system based on natural language processing
100: Natural language matching device based on morpheme analysis
110: pattern generation unit
111: morpheme analysis unit
112: Concept acquisition unit
113: pattern string creation unit
114: pattern definition unit
120: user input analysis unit
121: segmentation part
122: matching execution unit
200: electronic document writing device
300: electronic document processing device
400: electronic document retrieval device
500: navigation module
1: network
2: electronic document

Claims (12)

In the natural language matching device based on morpheme analysis,
A pattern generation unit that obtains concepts included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions, creates a pattern string, and generates a pattern based on the pattern string; And
A user input analysis unit that receives a user input, matches the pattern corresponding to the user input, and returns matching data,
Natural language matching device comprising a. The method of claim 1,
The pattern generation unit,
A morpheme analysis unit that performs a morpheme analysis on each of the plurality of sentences;
A concept acquisition unit that acquires a concept based on a result of morpheme analysis;
A pattern string creation unit for generating a pattern string including at least one of the concept or capture for extracting a specific phrase from the user input; And
It comprises a pattern definition unit for defining a pattern based on the pattern string,
Natural language matching device. The method of claim 2,
The result of the morpheme analysis,
A morpheme that analyzes each of the plurality of sentences in a morpheme unit and part of speech information associated with the analyzed morpheme,
The part-of-speech information is displayed based on a preset part-of-speech tag,
Natural language matching device. The method of claim 2,
The pattern character string creation unit is to create the pattern character string based on a concept set including at least one concept obtained by the concept acquisition unit,
Natural language matching device. The method of claim 2,
The above concept is
The natural language matching device, wherein a concept name encompassing words as elements of the concept exists and is defined based on the concept name or is defined in the form of an anonymous concept listing words as elements of the concept. The method of claim 2,
The user input analysis unit,
A segmentation unit for dividing the user input into a plurality of segments by cutting the user input into spaces; And
Comprising a matching performing unit that determines whether at least one segment matches the pattern and returns matching data when matching is successful,
Natural language matching device. The method of claim 6,
The matching performing unit,
Matching at least one segment and a word or matching the at least one segment and a concept, determining a pattern string including both the matched word and the matched concept, and returning matching data,
When the capture is included in the pattern string, the capture and the at least one segment are matched to extract a specific phrase in a user input associated with the capture,
The matching data includes at least one of data on the matched concept or a specific phrase in a user input associated with the capture,
Natural language matching device. In the electronic document control system based on natural language processing,
A natural language matching device that receives a user input, matches a pattern corresponding to the user input, returns matching data, and determines a user's intention contained in the user input based on the matching data;
An electronic document creation device that creates or corrects an electronic document based on an electronic form template according to the user's intention;
An electronic document processing device that performs approval, reception, or deletion processing for a previously created electronic document according to the intention of the user;
An electronic document search device that provides a list of electronic documents that meet a search condition included in a user input according to the user's intention; And
Electronic document control system based on natural language processing comprising a navigation module for controlling movement between a plurality of electronic documents according to the intention of the user. In the natural language matching method based on morpheme analysis,
Generating a pattern string by obtaining concepts included in the plurality of sentences through morpheme analysis of a plurality of sentences containing similar intentions, and generating a pattern based on the pattern string;
Receiving a user input; And
Matching the pattern corresponding to the user input and returning matching data,
Natural language matching method comprising a. The method of claim 9,
The step of generating the pattern,
Performing morpheme analysis for each of the plurality of sentences;
Obtaining a concept based on a result of morpheme analysis;
Creating a pattern string based on a concept set including at least one of the concepts or capture for extracting a specific phrase from the user input; And
Defining a pattern based on the pattern string,
That includes, natural language matching method. The method of claim 9,
Returning the matching data,
Dividing the user input into a plurality of segments by cutting the received user input into spaces; And
Determining whether the at least one segment matches the pattern and generating matching data when the matching is successful,
That includes, natural language matching method. The method of claim 11,
The step of generating the matching data,
Matching at least one segment and word or matching the at least one segment and concept;
Determining a pattern string including both the matched word and the matched concept;
Extracting a specific phrase in a user input associated with the capture by matching the capture and the at least one segment when the pattern string includes a capture; And
Generating the matching data including at least one of the data on the matched concept or a specific phrase in the user input associated with the capture,
That includes, natural language matching method.

Images