KR20070050305A - Method and system for indexing and retrieving in search engine and search service using query by initial sound - Google Patents

Abstract

The present invention relates to a first speech index and a search method provided by a search engine and a search service. More particularly, the present invention relates to a search interface that inputs a "first speech" search word made by only the first sound of each syllable constituting a search word. The first word index and search method using QBIS (Query By Initial Sound). To this end, in the present invention, the first spoken keyword is also indexed in addition to the conventional keyword at the time of indexing, and the search result can be output by comparing with the first spoken keyword. Alternatively, the first spoken search word may be converted into a conventional keyword and the user may select to output the search result. According to the first speech index and search method according to the present invention, by reducing the number of key-strokes when entering a search word, a user of a device that is inconvenient for the search word input means or a user who has difficulty in entering a search word or a user who cannot remember the search word correctly Can be done very conveniently.

First voice, search engine, QBIS, truncated search, query expansion

Description

METHOOD AND SYSTEM FOR INDEXING AND RETRIEVING IN SEARCH ENGINE AND SEARCH SERVICE USING QUERY BY INITIAL SOUND}

1 is a diagram illustrating a general indexing and searching process.

2 is a diagram illustrating a first speech indexing and searching method.

3 is a diagram illustrating a process of a search term expansion method, which is another implementation method of a first speech index and search method.

FIG. 4 is a block diagram illustrating a configuration of a first speech index and search system according to an exemplary embodiment of the present invention performing the process shown in FIG. 2.

FIG. 5 is a block diagram illustrating a configuration of a first speech index and search system according to another embodiment of the present invention performing the process shown in FIG. 3.

<Explanation of symbols for the main parts of the drawings>

400: First Speech Index and Search System

401: Keyword Extractor

402: First Speak Converter

403: index index component

404: search term input unit

405: document identification unit

406: search result provider

The present invention relates to a first speech index and a search method provided by a search engine and a search service. More particularly, the present invention relates to a search interface that inputs a "first speech" search word made by only the first sound of each syllable constituting a search word. The first word index and search method using QBIS (Query By Initial Sound).

Table 1 below shows various search terms input methods provided by commercial search engines and search services. In addition to the most basic keyword search terms, several short search terms are provided in commercial search engines and search services, such as abbreviation search terms found in truncated searches and acronym search terms found in synonym searches using thesaurus. It can be seen that.

Originally, truncated and synonymous searches produce more results than keyword keywords that only accurately enter the desired keywords. However, the accuracy of the search is slightly lower. However, if the search term is not remembered correctly or the search itself is more important than accuracy. For various reasons, such as the field of interest or the user's preference, these various short search terms are still used in commercial search.

Unlike conventional methods for inputting various search words, the present invention proposes a new type of search word input method called "Query By Initial Sound" (QBIS).

"First speech search" of the present invention is that literally, it is possible to specify a search word only by the initial consonant of the keyword to search. It is intended to make it easy to search for the desired document while greatly reducing the number of key-strokes when typing keywords.

In the case of English, as shown in Table 1, the initial search term has a low key-stroke effect, but Hangul uses the syllables combined with the elementary, middle, and final characters as unit letters. Even the initials don't reduce key-strokes much like English. Indeed, 'Major League Baseball' is 19 times, 'MLB' is 3 times, and the number of key-strokes is greatly reduced by 16% compared to the original word, while 'Information and Communication' is 13 times and 'Ministry of Information' is 8 It is 62% of circuit word and it does not decrease much compared to English. It may be generalized that this is caused by the syllable characteristics of Hangul because it is largely similar in other examples.

This leads to the inconvenience of requiring more key-strokes for the Hangul search users. Therefore, the present invention has devised a 'first speech' having a smaller number of keystrokes than the initial.

For users or input devices where a few more strokes of keyboard is not a problem, it is of course desirable to type a few more characters to get a more accurate search term. But for input devices or users who don't, "first voice" search may be a very useful search term interface and search method. For example, in a mobile terminal (smartphone, PDA phone, etc.) where the input interface is somewhat inconvenient compared to the desktop, it would be very convenient if the key-stroke can be greatly reduced to put a search word. Or, if the user feels difficulty in typing, for example, an elderly person unfamiliar with the computer keyboard or a disabled person who feels uncomfortable with the keyboard tapping operation, the first voice search according to the present invention will be very helpful. Or even if there is no problem with the input device and user typing, it can still provide a lot of convenience.

In order to achieve the above object, the first speech index and search method according to an embodiment of the present invention comprises the steps of extracting a keyword from the document through a predetermined keyword extractor; Performing first speech conversion on the extracted keyword to generate a first speech keyword corresponding to the extracted keyword; Constructing an Inverted Index table that stores the document identifier of the document corresponding to the first spoken keyword; Receiving a first search word from a user through a search word input window; Searching for a first spoken keyword corresponding to the input first spoken keyword from the inverted index table, and identifying a document identifier corresponding to the searched first spoken keyword; Generating a search result list by retrieving a document value using the identified document identifier; And providing the generated search result list to the user.

On the other hand, the first spoken index and search method according to another embodiment of the present invention comprises the steps of maintaining the entire keyword list including each keyword; Generating a first spoken keyword corresponding to each keyword by performing first speech conversion on each keyword; Storing the first spoken keyword and each keyword in correspondence with the entire keyword list; Receiving a first search word from a user through a search word input window; Searching one or more keywords corresponding to the input first search word from the entire keyword list and providing the searched one or more keywords to the user; Receiving a selection of a specific keyword from the one or more keywords provided by the user; And providing the user with a list of search results for the selected specific keyword.

General Indexing and Search Process

1 is a diagram illustrating a general indexing and searching process.

In step 101 a document is given. The ID of the given document is indicated in FIG. 1 as the tentative name D1.

In step 102, the original text of D1 is input to the keyword extractor. The keyword extractor is based on the morphological analysis technology, so that only core keywords can be extracted without unnecessary parts such as Hangul survey.

In step 103, a keyword is extracted. The extracted keywords are usually nouns, but may include verbs, adjectives, adverbs, etc. in some cases.

An inverted index table is constructed using the keywords extracted in step 104. If a document is given, the process of extracting a keyword is called an index. On the contrary, if a keyword is given, a process of finding a document is an inverse of the index.

The inverted index table includes a keyword area and a posting data area, so step 104 adds the keyword to the keyword area and adds a document identifier (ID) to its associated posting data area. The posting data area in this inverted index table can quickly identify which document a given keyword appears in, so it plays a very important role in the search. On the other hand, the inverted table may not be such a simple table in actual implementation, but may be structured to be suitable for searching a large number of keywords such as B + tree. In addition, the post location, the appearance frequency, the weight, and the like may be stored in the posting data area as well as the document identifier.

When the processes of steps 101 to 104 are repeatedly applied to a plurality of documents such as tentative names D2, D3, D4, and the like, the index is completed. Next, let's look at the search process.

In step 105 the user puts a search term in the search term input window.

In step 106, the entered search term is converted into a search expression and given to a search engine.

The keyword given in step 107 is searched in the inverted index table. If the inverted index is an array structure with an ordered array, a binary search may be used, and if the B + tree is a B + tree search, a search method may be employed. However, there may be differences in efficiency even though there are differences in search techniques, but there is no difference in results.

When the search finds the desired keyword in step 108, the posting data area of the keyword is referenced to identify in which document the keyword appeared.

In step 109, the document value is retrieved using the document identifier of the appearance document to generate a search result list.

In step 110, the generated search result list is displayed on the screen and provided to the user.

First speech  Index and Search Process

2 is a diagram illustrating a process of a first speech index and search method according to the present invention. The first spoken index and search method according to FIG. 2 may be performed in a predetermined first spoken index and search system.

Referring to Figure 2 looks at the flow of the index and search of the present invention. Since steps 201 to 210 of FIG. 2 are similar to steps 101 to 110 of FIG. 1, the description overlapping with FIG. 1 will be briefly described, and the characteristic process of FIG. 2 will be described in detail.

In step 201 a document is given. As in FIG. 1, the ID of a given document is denoted in FIG.

In step 202, the first spoken index and search system inputs the original text of D1 into the keyword extractor.

In step 203, the first speech index and search system extracts keywords from the original text through the keyword extractor. In this case, the keyword extractor may extract the keyword from the original text using morphological analysis, which is the same as described in FIG.

In step 204, the first spoken index and search system constructs an inverted index table with the extracted keywords. The inverted index table includes a keyword area and a posting data area, so in step 204 the first speech index and search system adds the keyword to the keyword area and assigns a document identifier (ID) to its associated posting data area. Add.

In step 204a, the first spoken index and search system converts the first spoken word for the keyword extracted in step 203, and generates a first spoken keyword corresponding to the extracted keyword in step 204b.

In this case, the first speech index and retrieval system first determines whether or not a code conversion is required depending on which charset the extracted keyword is encoded. In the past, we used codesets (sometimes referred to as KSSM combinations) that preserve per-phone encoding, but nowadays standards are established and very few such cases exist. Currently, the standard character set widely used is KSC5601-1987, KSC5601-1992, MSWIN949, UTF8, etc. These are all serial number codes for syllables. Thus, phoneme-specific code conversion can be said to be an essential process.

Next, the first spoken index and search system can take only the initial code from the converted code and convert it back to the original character set, thereby generating the first spoken keyword for the given keyword. In this case, keywords other than Korean keywords, that is, alphanumeric keywords can be ignored.

In step 204c, the first spoken index and search system constructs an Inverted Index table that stores the document identifier (ID) of the document corresponding to the first spoken keyword.

Once the first spoken keyword is also created, it has the same expression and structure as the general keyword. Thus, in step 204c, the first spoken index and search system adds the first spoken keyword to the keyword area of the inverted index table and adds the document identifier to the posting data area associated with the keyword area to which the first spoken keyword is added. By doing this, the inverted index table can be configured. In addition, the posting data area may store an appearance position, an appearance frequency, or a weight in a document associated with the first spoken keyword in addition to the document identifier.

When the processes of steps 201 to 204 and steps 204a to 204c are repeatedly applied to a plurality of documents such as tentatively D2, D3, D4, and the like, the index is completed. Next, let's look at the search process.

In step 205, the first spoken index and search system receives a first spoken search word from a user through a search word input window.

In step 206, the input search word is converted into a search expression and transmitted to a search engine.

In step 207, the first spoken index and search system searches the first spoken keyword corresponding to the input first spoken keyword from the inverted index table. If the index table is an ordered array structure, a binary search may be used, and if the B + tree is a B + tree search, the appropriate search scheme may be employed.

In addition, according to an embodiment of the present invention, the first voice keyword may be searched by at least one truncation search method among a right cut search, a frustrated search, and a left and right cut search according to the attribute of the input first search word. This will be described later.

In step 208, the first spoken index and search system identifies a document identifier corresponding to the retrieved first spoken keyword. In this case, the first spoken index and search system may identify the document identifier of the document in which the searched first spoken keyword appears with reference to the posting data area of the searched first spoken keyword.

In step 209, the first spoken index and search system takes a document value using the identified document identifier and generates a search result list.

In step 210, the first spoken index and search system displays the generated search result list on the screen and provides it to the user.

Truncation Search Method

Next, it will be described whether it is possible to apply truncation search and synonym search to the first spoken search word. In Table 2 below, it can be seen that various truncation searches are possible even with the first spoken search word. For example, a user who walks with a heavy bag in one hand and enters a search term on a smartphone with the other hand may find it helpful to shorten the search term even though the search results are a lot. In this case, the search term input method according to the truncation search method may be very effective.

As also described in Table 2, synonym searches are not well suited to first-speech searches. This is not because it is not impossible, but if you try to expand the synonym using only the first spoken condition, it can be extended to too many synonyms, which decreases the effectiveness.

Query expansion method

In the above description, a method of storing the first spoken keyword in an inverted index table at the time of searching for the first spoken word has been described. The present invention provides a search term expansion method in another configuration.

This method does not store the first spoken keyword in the inverse index table, but instead, presents the user with a list of keywords corresponding to the first spoken search word input by the user to select the first spoken search. For example, if a user enters 'ㅂ, ㅈ, ㅌ' as the first spoken query, the search engine will present a list of keywords, such as 'Byzantine', 'visitor', ... Select to search.

3 is a diagram illustrating a process of the search word expansion method, which is another implementation method of the first speech indexing and searching method according to the present invention. The first spoken index and search method according to FIG. 3 may also be performed in a predetermined first spoken index and search system.

First, the first speech indexing and retrieval system is a preliminary step for performing the steps 301 to 303 shown in FIG. 3, which includes: (1) maintaining an entire keyword list including each keyword, (2) Performing a first speech conversion for each keyword, generating a first speech keyword corresponding to each keyword, and (3) storing the first speech keyword and the respective keywords so as to correspond to the entire keyword list. can do.

The first speech index and search system may utilize the inverted index table shown in FIG. 1 as the entire keyword list, or alternatively, edit the entire keyword list separately. In addition, the first spoken index and search system converts the first spoken keywords to the secured keywords to generate the first spoken keyword corresponding to each keyword, and the first spoken keyword and the respective keywords correspond to the entire keyword list. An example of a complete list of keywords thus constructed is shown in FIG. 3.

In step 301, the first spoken index and search system receives a first spoken search word from a user through a search word input window. For example, as shown in FIG. 3, the first spoken index and search system may receive a first spoken search word '검색어 ㅈㅌ' from a user.

In step 302, the first spoken index and search system retrieves one or more keywords corresponding to the entered first spoken search word from the entire keyword list and provides the searched one or more keywords to the user. For example, as shown in FIG. 3, the first speech index and search system may search for keywords corresponding to 'ㅂ, ㅈ, ㅌ' in the entire keyword list and find 'Byzantine, visitor, ...' Can be presented back to the user. The first spoken index and search system may search the one or more keywords by at least one truncation search method among right cut search, frustrated search and left and right cut search according to the input first search word attribute.

In step 303, the first spoken index and search system receives a specific keyword from the one or more keywords provided by the user. For example, as shown in FIG. 3, the first speech index and search system may receive a keyword selected by the user from the extended search word list 'byzantine, visitor, ...'.

In the following process, the first speech index and search system may provide the user with a list of search results for the selected specific keyword, which is the same as steps 105 to 110 of FIG. Omit. As described above, the search term expansion method is the same as the general search method described with reference to FIG. 1 from the next step of selecting one of the extended search terms. Therefore, by adding such a search term expansion module to a general search engine, an existing first search term interface can be easily attached to an existing search engine.

The first speech indexing and retrieval method according to the present invention can be implemented in the form of program instructions that can be executed by various computer means and recorded in a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. The medium may be a transmission medium such as an optical or metal wire, a waveguide, or the like including a carrier wave for transmitting a signal specifying a program command, a data structure, or the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

FIG. 4 is a block diagram illustrating a configuration of a first speech index and search system according to an exemplary embodiment of the present invention performing the process shown in FIG. 2.

As shown in FIG. 4, the first speech index and search system 400 according to the present embodiment includes a keyword extractor 401, a first speech converter 402, an inverse index table forming unit 403, and a search term input unit 404. The document identification unit 405 and the search result providing unit 406 may be included.

The first speech index and search system 400 shown in FIG. 4 is for performing the process shown in FIG. 2, and a description of the overlapping description of FIG. 2 will be omitted.

The keyword extractor 401 extracts a keyword from the document.

The first spoken converter 402 performs first spoken conversion on the extracted keyword to generate a first spoken keyword corresponding to the extracted keyword.

An inverted index table constructing unit 403 constructs an inverted index table that stores the document identifier of the document in response to the first spoken keyword.

The search term inputter 404 receives a first speech search term from a user through a search term input window.

The document identification unit 405 searches the first spoken keyword corresponding to the input first spoken keyword from the inverted index table and identifies a document identifier corresponding to the searched first spoken keyword.

The search result providing unit 406 obtains a document value by using the identified document identifier, generates a search result list, and provides the generated search result list to the user.

FIG. 5 is a block diagram illustrating a configuration of a first speech index and search system according to another embodiment of the present invention performing the process shown in FIG. 3.

As shown in FIG. 5, the first spoken index and search system 500 according to the present embodiment includes a first spoken converter 501, an entire keyword list maintaining unit 502, a search term input unit 503, an extended keyword providing unit ( 504, a keyword selecting unit 505, and a search result providing unit 506.

The first spoken index and search system 500 shown in FIG. 5 is for performing the process shown in FIG. 3, and a description of the overlapping description of FIG. 3 will be briefly described below.

The first spoken converter 501 performs first spoken conversion on each keyword to generate a first spoken keyword corresponding to each keyword.

The entire keyword list holding unit 502 stores and maintains the first spoken keyword and the respective keywords so as to correspond to the entire keyword list.

The search word input unit 503 receives a first spoken search word from a user through a search word input window.

The extended keyword providing unit 504 searches one or more keywords corresponding to the input first search word from the entire keyword list, and provides the searched one or more keywords to the user.

The keyword selector 505 receives a specific keyword from the one or more keywords provided by the user.

The search result providing unit 506 provides the user with a list of search results for the selected specific keyword.

As described above, although the present invention has been described with reference to limited embodiments and drawings, the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

According to the first speech index and search method according to the present invention, a key-stroke when a user of a device (mobile terminal such as a mobile phone, a smart phone, a PDA phone, an intelligent home appliance such as a PVR, an interactive TV, etc.) that is inconvenient for a search word input unit enters a search word. The number of times can be made much smaller than before.

In addition, according to the first speech index and search method according to the present invention, even if a user remembers only a part or sound of a search word without thinking of an exact keyword, a search is possible.

In addition, according to the first speech index and search method according to the present invention, users (physical hand baggage, elderly, handicapped persons, etc.) who have a physical difficulty in entering a search word can be easily searched.

Claims (13)

In the first speech index and search method, Extracting a keyword from a document through a predetermined keyword extractor; Performing first speech conversion on the extracted keyword to generate a first speech keyword corresponding to the extracted keyword; Constructing an Inverted Index table that stores the document identifier of the document corresponding to the first spoken keyword; Receiving a first search word from a user through a search word input window; Searching for a first spoken keyword corresponding to the input first spoken keyword from the inverted index table, and identifying a document identifier corresponding to the searched first spoken keyword; Generating a search result list by retrieving a document value using the identified document identifier; And Providing the generated list of search results to the user First speech index and search method comprising a. The method of claim 1, The step of generating a first speech keyword corresponding to the extracted keyword by performing a first speech conversion on the extracted keyword, Determining whether a code conversion is necessary according to which charset the extracted keyword is encoded; Performing transcoding for each phoneme if transcoding is necessary; And Taking only the initial code from the converted code and converting it back to the original character set to generate the first spoken keyword corresponding to the extracted keyword First speech index and search method comprising a. The method of claim 1, The step of extracting a keyword from a document through a predetermined keyword extractor, Inputting the original text of the document into the keyword extractor; And Extracting the keyword from the original text using morphological analysis in the keyword extractor First speech index and search method comprising a. The method of claim 1, The inverted index table includes a keyword area and a posting data area, The step of configuring an inverted index table for storing the document identifier of the document corresponding to the first spoken keyword, Adding the first spoken keyword to the keyword area and adding the document identifier to a posting data area associated with the keyword area to which the first spoken keyword is added; First speech index and search method comprising a. The method of claim 4, wherein And a appearing position, a appearing frequency, or a weight in a document associated with the first spoken keyword in addition to the document identifier. The method of claim 4, wherein The step of identifying a document identifier corresponding to the searched first spoken keyword, And a document identifier of a document in which the searched first speech keyword appears, by referring to the posting data area of the searched first speech keyword. The method of claim 1, The step of searching for the first spoken keyword corresponding to the input first search word from the inverted index table, The first speech index and the search method, characterized in that for searching the first speech keyword by at least one truncation search method of the right cutting search, frustration search, left and right cut search according to the attribute of the input first speech search word. In the first speech index and search method, Maintaining a complete list of keywords including each keyword; Generating a first spoken keyword corresponding to each keyword by performing first speech conversion on each keyword; Storing the first spoken keyword and each keyword in correspondence with the entire keyword list; Receiving a first search word from a user through a search word input window; Searching one or more keywords corresponding to the input first search word from the entire keyword list and providing the searched one or more keywords to the user; Receiving a selection of a specific keyword from the one or more keywords provided by the user; And Providing the user with a list of search results for the selected specific keyword First speech index and search method comprising a. The method of claim 8, The step of performing the first speech conversion for each keyword, generating the first speech keyword corresponding to each keyword, Determining whether a code conversion is required according to which charset each keyword is encoded; Performing transcoding for each phoneme if transcoding is necessary; And Taking only the initial code from the converted code and converting it back to the original character set to generate the first spoken keyword corresponding to each keyword; First speech index and search method comprising a. The method of claim 8, The step of searching for one or more keywords corresponding to the input first search word from the entire keyword list, The first voice index and the search method, characterized in that for searching for the one or more keywords by at least one truncation search method of the right cutting search, frustration search, left and right cut search according to the attribute of the input first search word. A computer-readable recording medium in which a program for executing the method of any one of claims 1 to 10 is recorded. In the first speech index and search system, A keyword extractor for extracting keywords from the document; A first speech converter configured to perform first speech conversion on the extracted keyword to generate a first speech keyword corresponding to the extracted keyword; An inverted index table constructing unit for constructing an inverted index table for storing a document identifier of the document corresponding to the first spoken keyword; A search term input unit which receives a first search term from a user through a search term input window; A document identification unit searching for a first spoken keyword corresponding to the input first spoken keyword from the inverted index table and identifying a document identifier corresponding to the searched first spoken keyword; And A search result providing unit which retrieves a document value using the identified document identifier to generate a search result list and provides the generated search result list to the user First speech index and search system comprising a. In the first speech index and search system, A first speech converter for performing a first speech conversion for each keyword to generate a first speech keyword corresponding to each keyword; An entire keyword list holding unit for storing and storing the first spoken keyword and the respective keywords corresponding to the entire keyword list; A search term input unit which receives a first search term from a user through a search term input window; An extended keyword providing unit for searching one or more keywords corresponding to the input first search word from the entire keyword list and providing the searched one or more keywords to the user; A keyword selector configured to select a specific keyword from the one or more keywords provided by the user; And A search result provider for providing the user with a list of search results for the selected specific keyword First speech index and search system comprising a.

Images