KR20160111180A

KR20160111180A - Apparatus and Method for completion of query based on natural language

Info

Publication number: KR20160111180A
Application number: KR1020150035994A
Authority: KR
Inventors: 서형일; 주인학; 지형근
Original assignee: 한국전자통신연구원
Priority date: 2015-03-16
Filing date: 2015-03-16
Publication date: 2016-09-26

Abstract

A natural-language type-based query completion device, method, and retrieval system are disclosed. According to one aspect of the present invention, there is provided a method for completing a natural-language form-based query, comprising the steps of: determining a central keyword; searching for a question formation candidate to be connected to a tail of the determined central keyword according to a natural language form; And selecting one of the forming candidates to form a question.

Description

Technical Field [0001] The present invention relates to a method,

The present invention relates to a query completion technique, and more particularly, to a query completion device, a method, and a retrieval system based on a natural language type.

Information retrieval services include keyword retrieval, natural language retrieval, and browse retrieval.

Keyword search method is the most representative method provided by general search portal site. However, since it is difficult to clearly express the intention of a user with a keyword alone, the precision and the recall ratio of the search result are low.

Natural language search is the most natural search method for people. However, because the structure of natural language is so complex, it is very difficult for computers to understand natural language. Therefore, the precision of the search result through the natural language search method is low at the current technology level. In addition, since the natural language search can be performed only in a pre-programmed search, the user must know in advance which query words should be used for searching according to his / her intention.

The browse search method is a method of searching for a predetermined category. The browse search method has the advantage of high precision but it does not reflect the change rapidly, so the recall ratio is low in a rapidly changing environment. In addition, the flexibility is low in that the user can browse only through a predefined path, and the efficiency of the search is low because the entire information structure must be learned in the browsing process.

A method, apparatus, and system for query completion based on natural language that can expand a keyword sequentially based on a natural language format to complete a natural language query.

According to one aspect of the present invention, there is provided a method for completing a natural-language form-based query, comprising the steps of: determining a central keyword; searching for a question formation candidate to be connected to a tail of the determined central keyword according to a natural language form; And selecting one of the forming candidates to form a question.

The step of determining the center keyword may include searching a central keyword candidate using the keyword auto-completion function based on the input first keyword, and selecting one of the searched central keyword candidates as a center keyword have.

The step of searching for the query formation candidate may retrieve the query formation candidate matching the determined center keyword in a prebuilt database.

A method for completing a natural-language type-based question includes the steps of: determining whether the formed question is completed with a natural-language question; searching the question-forming candidate if the natural-language question is not completed as a result of the determination; Wherein the step of searching for a question formation candidate may search for an additional question formation candidate to be connected to the tail of the formed question according to a natural language format.

A natural language form-based question completion apparatus according to another aspect includes a central keyword determination unit for determining a central keyword, a question formation candidate for searching a question formation candidate to be connected to the end of the determined central keyword according to a natural language form, And a question generating unit for selecting one of the retrieved question formation candidates and forming a question.

The central keyword determination unit searches the central keyword candidate using the keyword automatic completion function based on the input first keyword, and selects one of the searched central keyword candidates as the center keyword.

Wherein the natural language type based query completion device further includes a question formation candidate database for storing question formation candidates matched with respective central keywords, and the question formation candidate retrieval section includes a question formation candidate matching the determined central keyword in the database You can search.

The question formation candidate search unit may search for an additional question formation candidate to be connected to the end of the formed question according to the natural language form when the formed question is not completed with the natural language question.

According to another aspect of the present invention, there is provided a terminal comprising: an input unit for receiving a keyword; a question completion unit for confirming a central keyword based on the input keyword and expanding the determined central keyword based on a natural language format, A communication unit for transmitting the natural language query to the server and receiving a search result for the natural language query completed by the server, a user interface for completing the natural language query, and a display unit for displaying the search result received from the server .

The query completion unit includes a center keyword determination unit for determining a center keyword based on the input keyword and a query formation candidate search unit for searching for a question formation candidate to be connected to the end of the determined center keyword according to the natural language format, And a question formation unit for selecting one of the searched question formation candidates to form a question.

The central keyword determination unit searches the central keyword candidate using the keyword automatic completion function based on the input keyword, and selects one of the searched central keyword candidates as the center keyword.

The user interface for completing the natural language query can be divided into a part for displaying the inputted keyword, a part for displaying the question formed so far, and a part for displaying the searched question formation candidate.

The terminal may be a mobile terminal.

The reliability of the search result can be increased by sequentially expanding the keyword based on the natural language type and completing the natural language query and using it for the search.

In addition, it is possible to obtain high precision and recall ratio by securing high flexibility using a natural language expansion method and limiting the search word to a limited range.

1 is a diagram showing an embodiment of a retrieval system.
2 is a detailed configuration diagram of the natural language-based retrieval system 100 of FIG.
3 is a detailed configuration diagram of the question completion unit 112 of FIG.
FIG. 4 is an exemplary view of information stored in the question formation candidate database 320 for completion of a Korean noun phrase query.
5 is an illustration of a completed Korean noun phrase query based on the information of FIG.
6 is a diagram showing an embodiment of a UI displayed on the display unit 114. As shown in FIG.
7 is a diagram showing another embodiment of the UI displayed on the display unit 114. In Fig.
8 is an example of completing an English noun phrase question.
9 is a view showing an embodiment of a UI displayed on the display unit 114 for performing an in-result re-search.
10 is a diagram showing another embodiment of the retrieval system.
11 is a flowchart showing an embodiment of a method for completing a natural language type based query.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. In addition, the terms described below are defined in consideration of the functions of the present invention, which may vary depending on the intention of the user, the operator, or the custom. Therefore, the definition should be based on the contents throughout this specification.

1 is a diagram showing an embodiment of a retrieval system.

Referring to FIG. 1, a search system 100 may include a terminal 110 and a server 120.

The terminal 110 may include a mobile terminal and a fixed terminal, which are capable of completing a natural language query by sequentially expanding a keyword inputted from a user based on a natural language format. At this time, the mobile terminal may be a wearable device such as a mobile phone, a sprite phone, a tablet PC, a laptop computer, a digital broadcast terminal, a PDA (Personal Digital Assistant), a PMP (Portable Multimedia Player) , And Internet of Things (IoT) devices, and the fixed terminal may include a digital TV, a smart TV, a desktop computer, and the like.

The server 120 may receive a natural language query from the terminal 110 and perform a search for a natural language query. At this time, the server 120 may divide the natural language query into each element and utilize the natural language query as an input of the search.

Hereinafter, a natural language-based retrieval system 100 according to an embodiment will be described in detail with reference to FIG.

2 is a detailed configuration diagram of the natural language-based retrieval system 100 of FIG.

2, the terminal 110 may include an input unit 111, a query completion unit 112, a communication unit 113, and a display unit 114.

The input unit 111 may include a plurality of input keys and function keys for receiving numeric or character information and setting various functions, as means for receiving commands, selections, data, information, etc. of the user.

For example, the input unit 111 may include a key pad, a dome switch, a touch pad (resistive / capacitive), a jog wheel, A jog switch, an H / W button, and the like. Particularly, when the touch pad has a mutual layer structure with the display unit 150, it can be called a touch screen.

The query completion unit 112 may complete the natural language query based on the first keyword received through the input unit 110. [ For example, the query completion unit 112 can determine a central keyword based on the initial keyword, and expand the determined central keyword based on the natural language type to complete the natural language query.

In addition, the query completion unit 112 can complete a new natural language query based on the search result of the completed natural language query so that the terminal 110 can perform the in-result rediscovery function. For example, the query completion unit 112 can confirm a search result for a completed natural language query as a central keyword, and expand a newly determined central keyword based on a natural language format to complete a natural language query.

The detailed description of the question completion unit 112 will be described later with reference to FIG.

The communication unit 113 can transmit and receive data to and from the server 120 via a wired or wireless network. Here, the wired or wireless network may include all communication networks capable of Internet communication using various protocols such as a mobile communication network, a wired or wireless public network, and a private network.

The communication unit 110 may transmit the completed natural language query to the server 120 in the query completion unit 112 and receive the search result corresponding to the natural language query from the server 120. [

The display unit 114 can display information processed by the terminal 110. [ The display unit 114 may display a UI (User Interface) or a GUI (Graphic User Interface) for natural language question formation / completion and a search result received from the server 120 through the communication unit 113.

The display unit 114 may be a liquid crystal display (LCD), a thin film transistor liquid crystal display (TFT-LCD), an organic light emitting diode (OLED), a flexible display, , And a three-dimensional display (3D display).

In addition, two or more display units 114 may exist according to the implementation form of the terminal 110. [ For example, in the terminal 110, the plurality of display portions 114 may be spaced apart from one another or disposed integrally with each other, and may be disposed on different surfaces.

The display unit 114 may include a touch screen and a touch screen. The display unit 114 may be used as an input device in addition to the output device.

The server 120 may include a communication unit 121, a search unit 122, and a search object DB 123.

The communication unit 121 can transmit and receive data to and from the terminal 110 via a wired or wireless network. Here, the wired or wireless network may include all communication networks capable of Internet communication using various protocols such as a mobile communication network, a wired or wireless public network, and a private network.

The communication unit 121 can receive the natural language query completed by the query completion unit 112 from the communication unit 113 of the terminal 110 and transmit the search result corresponding to the natural language query to the communication unit 113. [

The search unit 122 can search the search object DB 123 based on the natural language query received from the terminal 110. [ According to one embodiment, the search unit 122 may divide the natural language query into elements and use the natural language query as an input for the search. For example, when the natural language question received from the terminal 110 is a noun phrase "What is the day of the week? &Quot;, the retrieval unit 122 sets the natural language question" "), A clause predicate (" to do "), and a noun (" day of the week? "). In this case, the tentative phrase descriptor ("to do") is used to determine the type of function, and the object type ("photographing") and the object type (event or action) And a filter value, and a noun ("what day of the week?") Can be used to determine the grouping direction of the result value.

The search target DB 123 can store various information to be searched. According to one embodiment, the search object DB 123 may store information about various web pages collected on the web. The web page information may be stored in a plurality of search lists according to the own criteria of the server 120. Each search listing may include a title, a URL, a description of the web page, an index word, and the like. Here, the title is the name given to the web page, and the URL is the network address of the web page.

3 is a detailed configuration diagram of the question completion unit 112 of FIG.

3, the query completion unit 112 may include a central keyword determination unit 310, a question formation candidate database 320, a question formation candidate search unit 330, and a question formation unit 340 .

The center keyword determination unit 310 can determine the center keyword that is the basis of the natural language question formation. According to an exemplary embodiment, the central keyword determination unit 310 may perform a keyword auto-completion function to determine a central keyword. Here, the keyword auto-completion function may mean a function of providing various keyword candidates to the user including the input character even if only a part of the keyword is inputted so that the user can express the contents to be searched as much as possible. At this time, the keyword auto-completion function can be provided based on analyzing the usage pattern of the user who uses the search or the input pattern of various keywords, or by reflecting the issue at the time of inputting a part of the keyword, May be provided.

For example, when the keyword 'photo' is input from the user through the input unit 111, the central keyword determination unit 310 determines the center of the 'photo shooting', 'photograph shop', and ' Keyword candidates can be provided to the user. In addition, the center keyword determination unit 310 can determine 'photo shoot' selected by the user as a center keyword when the user selects 'photograph shooting' among the central keyword candidates.

The query formation candidate database 320 may store various query formation candidates matched with the central keyword and various additional query formation candidates matched to each query formation candidate.

The query formation candidate search unit 330 may search for a query formation candidate to form a query by being connected to the tail of the central keyword based on a natural language format. For example, the question formation candidate searching unit 320 may search for the question formation candidates matching the central keyword in the query formation candidate database 320. [

The question formation unit 340 may select one of the query formation candidates retrieved in the query formation candidate retrieval unit 330 and form a natural language query by connecting the selected query formation candidate to the end of the central keyword. For example, the question formation unit 340 selects one of the question formation candidates retrieved in the question formation candidate retrieval unit 330 according to the user's selection command input through the input unit 111, You can connect to the end.

According to one embodiment, the question completion unit 112 may repeat the question formation candidate search and the natural language question formation until the formed question is completed with the natural language question. At this time, the question formation candidate searching unit 330 searches for an additional question formation candidate to be connected to the end of the question formed in the question formation unit 340, and the question formation unit 340 selects one of the additional question formation candidates, You can link to the end of the formed question.

Hereinafter, the natural language query completion process will be described in detail with reference to FIGS. 4 and 5. FIG.

FIG. 4 is an example of information stored in the question formation candidate database 320 for completion of a Korean noun phrase query, and FIG. 5 is an example of a Korean noun phrase query completed based on the information in FIG. Here, the Korean noun phrase question can consist of a noun that receives the idiom object, the idiomatic predicate, and the idiom of the idiom.

3 to 5, when the user inputs the first keyword "PHOTO ", the central keyword determination unit 310 performs a keyword auto-completion function based on the first keyword" PHOTO " Photographing ", and" our photograph shop "are presented to the user as a key keyword candidate.

When the user selects "photographed" among the central keyword candidates, the central keyword determination unit 310 determines the "photographing"

The question formation candidate search unit 330 selects the "photo shoot" determined as the central keyword as the object of the casting balloon, searches the question formation candidate database 320, and outputs the "casting quotation predicate matching with" photographing " And "concentrated" are extracted as question formation candidates and presented to users. At this time, the pendulum phrase predicate can be changed according to the type of the object of the pendulum.

When the user selects "to do" among the question formation candidates, the question formation unit 340 forms a question "take a picture" by concatenating "to"

In order to complete the noun phrase natural language query, the question formation candidate retrieval unit 330 searches the question formation candidate database 320 to obtain a noun phrase " The nouns that match "to do" and "place type" are candidates for additional questions, such as "time zone?", "Day of the week?", "Month?", "Season", "place name" And presents it to the user. At this time, nouns may vary according to the adjective clause.

If the user selects "What day of the week?" Among the additional question formation candidates, the question formation unit 340 connects the "What day of the week?" "To complete the natural language query as shown in Fig.

6 is a diagram showing an embodiment of a UI displayed on the display unit 114. As shown in FIG. 6 is an example of a mobile interface environment.

6, the UI displayed on the display unit 114 includes a portion 610 for receiving and displaying text information such as an initial keyword, a portion 620 for displaying a question formed so far, And a portion 620 for displaying a question formation candidate to be connected.

7 is a diagram showing another embodiment of the UI displayed on the display unit 114. In Fig. FIG. 7 is an example of an interface environment larger than that of FIG.

Since the interface screen is a small mobile environment, only one step is displayed on the screen at a time, but FIG. 7 is an interface environment larger than that of FIG. 6, so that all the steps can be displayed on one screen.

8 is an example of completing an English noun phrase question. English noun phrases consist of nouns, which are the core of the question, and related clauses, which modify the nouns. At this time, the related clause comes from the subject and the verb of the related company, followed by the object or a noun that supplements the corresponding verb. Because English is the opposite of Korean and word order, the noun at the core of the question is at the end of the question in the case of Korean, but in English, the noun at the beginning of the question is in the case of the question.

Therefore, in the case of an English question, it is possible to search using only the center keyword "Place Name " determined by the keyword automatic completion function based on the initial keyword" Place " inputted by the user. , Finish Here button) (810) to search by "Place Name" alone.

9 is a view showing an embodiment of a UI displayed on the display unit 114 for performing an in-result re-search.

As shown in FIG. 9, in order for the terminal 110 to perform the re-search in the result, the question completion unit 112 newly generates a new natural language query based on the search result of the " Natural language questions can be completed.

10 is a diagram showing another embodiment of the retrieval system.

Referring to FIGS. 2 and 10, the search system 1000 of FIG. 10 differs from the search system 100 of FIG. 2 in that the query completion unit 112 is included in the server 1020, . In this case, the terminal 1010 only transmits the user input to the server 1020 and displays the information received from the server 1020, and the formation and completion of the inquiry can all be done in the server 1020.

11 is a flowchart showing an embodiment of a method for completing a natural language type based query.

Referring to FIG. 11, in the natural language form-based question completion method, a central keyword that is a basis of natural language question formation is determined (1110). For example, the query completion unit 112 searches for a central keyword candidate using the keyword auto-completion function based on the input first keyword, and selects one of the searched central keyword candidates as a center keyword. Here, the keyword auto-completion function may mean a function of providing various keyword candidates to the user including the input character even if only a part of the keyword is inputted so that the user can express the contents to be searched as much as possible. At this time, the keyword auto-completion function can be provided based on analyzing the usage pattern of the user who uses the search or the input pattern of various keywords, or by reflecting the issue at the time of inputting a part of the keyword, May be provided.

Then, a query formation candidate to be connected to the tail of the central keyword according to the natural language format is searched for (step 1120). For example, the query completion unit 112 may search the question formation candidate database 320 for a question formation candidate matching the central keyword.

Thereafter, one of the searched candidate questions is selected to form a question (1130). For example, the question completion unit 112 may select one of the searched question formation candidates and connect the selected question formation candidate to the tail of the central keyword according to the inputted selection instruction of the user.

Thereafter, it is determined whether the formed question is completed as a natural language question (step 1140). If the question is not completed with the natural language question, the process returns to step 1120 and is connected to the end of the formed question to search for an additional question formation candidate to form the question.

One aspect of the present invention may be embodied as computer readable code on a computer readable recording medium. The code and code segments implementing the above program can be easily deduced by a computer programmer in the field. A computer-readable recording medium may include any type of recording device that stores data that can be read by a computer system. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, and the like. In addition, the computer-readable recording medium may be distributed to networked computer systems and written and executed in computer readable code in a distributed manner.

The present invention has been described with reference to the preferred embodiments. 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 invention as defined by the appended claims. Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be construed to include various embodiments within the scope of the claims.

100: Search system
110: terminal
111:
112:
113:
114:
120: Server
121:
122:
123: Search target DB
310: center keyword determination unit
320: Question formation candidate DB
330: query formation candidate search unit
340:

Claims

Determining a center keyword;
Searching for a question formation candidate to be connected to a tail of the determined center keyword according to a natural language format to form a question;
Selecting one of the searched question formation candidates to form a question; A method of completion of a natural-language type-based question.

The method according to claim 1,
Wherein the step of determining the central keyword comprises:
Searching for a central keyword candidate using an automatic keyword completion function based on an input first keyword; And
Selecting one of the searched central keyword candidates as a center keyword; A method of completion of a natural-language type-based question.

The method according to claim 1,
The step of searching for the question formation candidate comprises:
And searching for a question formation candidate matching the determined center keyword in a prebuilt database.

The method according to claim 1,
Determining whether the formed question is completed with a natural language question; And
Repeating the step of searching for the question formation candidate and the step of forming the question if the natural language query is not completed as a result of the determination. Further comprising:
Wherein the step of searching for the question formation candidate searches for an additional question formation candidate to be connected to the tail of the formed question according to a natural language format.

A central keyword determination unit for determining a central keyword;
A query formation candidate search unit connected to the end of the determined center keyword according to a natural language format to search for a query formation candidate to form a query;
A question formation unit for selecting one of the searched question formation candidates to form a question; Based query completion device.

6. The method of claim 5,
Wherein the central keyword determination unit searches for a central keyword candidate using the keyword automatic completion function based on the input first keyword and selects one of the searched central keyword candidates as a center keyword.

6. The method of claim 5,
A query formation candidate database for storing query formation candidates matched to respective center keywords; Further comprising:
Wherein the question formation candidate search unit searches the database for a question formation candidate matched with the determined center keyword.

6. The method of claim 5,
Wherein the question formation candidate search unit comprises:
And if the formed question is not completed as a natural language question, searches for an additional question formation candidate to be connected to the end of the formed question according to the natural language form.

An input unit for receiving a keyword;
A query completion unit for determining a central keyword based on the input keyword and expanding the determined central keyword based on a natural language format to complete a natural language query;
A communication unit for transmitting the completed natural language query to the server and receiving a search result for the natural language query completed by the server; And
A display unit for displaying a user interface for completing a natural language query and a search result received from the server; .

10. The method of claim 9,
The question completion unit,
A center keyword determiner for determining a center keyword based on the input keyword;
A query formation candidate search unit connected to the end of the determined center keyword according to a natural language format to search for a query formation candidate to form a query; And
A question formation unit for selecting one of the searched question formation candidates to form a question; .

11. The method of claim 10,
Wherein the center keyword determination unit searches for a central keyword candidate using the keyword automatic completion function based on the input keyword and selects one of the searched central keyword candidates as a center keyword.

11. The method of claim 10,
The user interface for completing the natural language query includes:
A terminal which is divided into a part for displaying the inputted keyword, a part for displaying the question formed up to now, and a part for displaying the searched candidate for question formation.

10. The method of claim 9,
Wherein the terminal is a mobile terminal.