KR102083017B1 - Method and system for analyzing social review of place - Google Patents

Abstract

The present invention relates to a method and system for analyzing social reviews of a place. A method for analyzing a social review of a place according to an embodiment of the present invention includes receiving search keywords and category information related to the place from a user; Collecting a social review of the search keyword; Performing a morpheme analysis of the social review by using a word dictionary for each application area corresponding to category information input from a user among word dictionary for each application area provided for each category so that different words are included for each category; And extracting and outputting analysis information for the social review using the information derived by the morpheme analysis.

Description

Method and system for analyzing social review of place}

The present invention relates to a method and system for analyzing social reviews of a place for providing analysis information about a place such as a restaurant, a lodging facility, and a travel spot using social big data.

With the emergence of various social network services in recent years, users have an increased opportunity to easily share their experiences or opinions with others, and accordingly, attempts have been made to utilize social network services as a marketing means.

In particular, recently, reviews and evaluations of restaurants, lodging establishments, and tourist attractions are shared through portal sites or SNS sites. Online reviews, evaluations, etc. of these places are important criteria for consumers to decide whether to select the place. Is becoming.

Furthermore, when establishing the marketing strategy of the relevant place or analyzing commercial areas, the necessity of using analysis information for social reviews is gradually emerging, and related technologies have been developed and proposed. (See Patent Document 1)

To analyze social reviews collected online, morphological analysis of review data must be preceded. Such morphological analysis recognizes the type of words (nouns, surveys, verbs, etc.), extracts valuable nouns as keywords Convert a word to a basic form (standard form).

The morpheme analysis is basically performed by searching a word dictionary containing words necessary for analysis. This process is performed by searching the entire Hangul dictionary for each word of the review, and if the searched word is an adjective or a verb, all possible forms of utilization must be checked / processed, which consumes a lot of time and computing resources. For example, assuming a commercial analysis service that updates national commercial analysis information on a monthly basis, it takes about 1 minute to process 500 reviews, and assuming an average of 500 reviews per place, 35000 nationwide It takes about 15 days to process a review of a dog's place. According to this, if the time required for collecting reviews is included, monthly updates are virtually impossible.

Meanwhile, a step of extracting valid data from a review that has undergone morphological analysis is performed. In the existing method of extracting valid data, a predetermined positive expression (for example, after the collected social review is regarded as having little relation with a given search keyword) , 'Delicious', 'Delicious', etc.) is used to extract only the reviews that contain it, and this method has a problem in that the rate of adoption as effective data is remarkably low. As described above, when the number of reviews serving as a population of statistics is small, accuracy and reliability of the calculation result are deteriorated, and a problem is vulnerable to viral marketing.

Registered Patent Publication No. 10-1715737 (2017,03.07)

The present invention is to solve the above problems, as well as to provide an improved speed than the prior art, and to provide a method and system for analyzing social reviews of a place capable of improving analysis accuracy.

The technical problems to be achieved by the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned will be clearly understood by those skilled in the art from the following description. Will be able to.

According to an embodiment of the present invention, receiving search keyword and category information about a place from a user; Collecting a social review of the search keyword; Performing a morpheme analysis of the social review by using a word dictionary for each application area corresponding to category information input from a user among word dictionary for each application area provided for each category so that different words are included for each category; And extracting and outputting analysis information for the social review using the information derived by the morpheme analysis.

In addition, the category information may include one or more of restaurants, cafes, lodging establishments, and tourist attractions.

Further, the analysis information for the social review may include one or more of a related keyword, a related image, a popularity index, and recognition information.

In addition, the method for analyzing the social review may further include extracting valid data for outputting analysis information by excluding all reviews having a low noise pattern after considering all collected social reviews as relevant. have.

In addition, the step of extracting the effective data may be performed by searching for a pattern dictionary for each application area including different noise patterns for each category.

In addition, the social review analysis method further includes the step of translating the analysis information on the social review into another language, and the translation step is performed by searching for a dictionary of proper nouns separately provided to include translation information about the proper noun. Performing a translation for the noun; It may include the step of performing the translation for the remaining part except the proper noun through the online translator.

Meanwhile, according to another embodiment of the present invention, a social review collection unit that receives search keywords and category information regarding a place from a user and collects social reviews of the search keywords; A morpheme analysis unit that performs a morpheme analysis for the social review; And an analysis unit for extracting and outputting analysis information for the social review using the information derived from the morpheme analysis, wherein the morpheme analysis unit is provided for each category, and different words for each category are provided. A word dictionary for a plurality of application areas included; And a morpheme analyzer that performs a morpheme analysis on the social review by using a word dictionary for each application region corresponding to category information input from a user among the word dictionary for each of the plurality of application regions. An analysis system can be provided.

In addition, the morpheme analysis unit may further include a word dictionary learner that derives words included in the word dictionary for each application area through machine learning of test data.

In addition, the social review analysis system, after considering all the collected social reviews as relevant, excludes the reviews that include the low noise pattern and extracts effective data for output of analysis information. It may further include.

Further, the noise filter unit may perform filtering by searching a pattern dictionary for each application area including different noise patterns for each category.

Further, the word dictionary and the pattern dictionary for each application area may be configured to be modularized for each category.

In addition, the social review analysis system further includes a translation unit that translates the analysis information on the social review into another language, and the translation unit includes: a proper noun dictionary including translation information about a proper noun; And the proper noun part of the analysis information may include a translator that performs translation using the proper noun dictionary, and performs translation through an online translator for the remaining parts except the proper noun. Here, the proper noun dictionary may be provided for each of a plurality of languages.

According to an embodiment of the present invention, by preparing a word dictionary for each application area for each category, and performing a morpheme analysis for a social review using the word dictionary for each application area corresponding to the category information received from the user, It can greatly improve the processing speed of the system, and has the effect of reducing the consumption of computing resources.

In addition, as a method of extracting valid data among the collected social reviews, by analyzing all collected social reviews as relevant, by using a positive noise filtering method that excludes reviews that include a noise pattern with a low correlation. The accuracy and reliability of the results can be improved.

In addition, it is possible to expand the use layer of the analysis information to foreigners by providing translation information for commercial analysis information, and perform the translation of the original noun through a separate noun dictionary prepared first when translating, and then use the remaining translations as an online translator. By performing, there is an advantage that can minimize the mistranslation that may occur during translation through an online translator.

Further, by configuring the word dictionary and pattern dictionary for each application area to be modularized by category, there is an advantage of easy addition of categories and expansion of application fields.

1 is a block diagram of a social review analysis system of a place according to an embodiment of the present invention.
2 is a flow chart illustrating a method for analyzing social reviews of a place according to an embodiment of the present invention.
3 is a diagram illustrating an example of an input page for receiving search keyword and category information related to the present invention.
4 is a diagram illustrating an example of outputting valid data after noise filtering for a social review related to the present invention.
5 is a diagram illustrating an example of an output page that outputs analysis information for a social review related to the present invention.

The present invention can be applied to a variety of transformations and may have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all conversions, equivalents, and substitutes included in the spirit and scope of the present invention. In the description of the present invention, when it is determined that a detailed description of known technologies related to the present invention may obscure the subject matter of the present invention, the detailed description will be omitted.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from other components.

Terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprises" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts or combinations thereof described in the specification, one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance.

Hereinafter, an embodiment of a method and system for analyzing social reviews of a place according to the present invention will be described in detail with reference to the accompanying drawings. In describing with reference to the accompanying drawings, identical or corresponding components have the same reference numbers. It will be given and redundant description will be omitted.

1 is a block diagram of a place's social review analysis system according to an embodiment of the present invention.

Referring to FIG. 1, the social review analysis system according to the present embodiment includes a social review collection unit 10, a morpheme analysis unit 20, and an analysis unit 30.

The social review collection unit 10 receives search keywords and category information regarding a place (restaurant, cafe, accommodation, tourist attraction, etc.) from a user. Here, the search keyword may be a place name or an address name of the place, and the category information may be a 'restaurant', a 'cafe', a 'hotel,' or a tourist attraction. The user may input the corresponding information through a user client (smartphone, computer, etc.) connected online with the social review collection unit 10.

The social review collection unit 10 collects social reviews of search keywords received from a user. The social review collection unit 10 collects social reviews from a predetermined social site 100, and the social site 100 includes portal sites such as Naver, Daum, and SNS sites such as Facebook, Twitter, and Instagram. And the like. Social reviews include posts such as blogs, cafes, and social media feeds, and can include images in addition to text. Such social review data can be collected through an open search API provided by portals and SNS sites, and the collected social reviews can be stored in a separately prepared social review database 15.

The morpheme analysis unit 20 performs morpheme analysis for social reviews. The morpheme analysis unit 20 recognizes types of words (nouns, surveys, verbs, etc.) constituting social reviews, extracts valuable nouns as keywords, and converts each word into a basic form (standard form).

The morpheme analysis unit 20 includes a word dictionary 21 for each application area and a morpheme analyzer 22.

A plurality of word dictionaries 21 for each application area are provided for each category, and different words are included for each category. For example, the words frequently mentioned in social reviews are different for restaurants, cafes, lodging establishments, and tourist attractions, and by using these matters, the dictionary of words for each application service is separately produced to reduce the size of the dictionary. The word dictionary 21 for each application area may be composed mainly of words having a main meaning for each application area.

The morpheme analyzer 22 performs morpheme analysis for a social review by using the word dictionary 21 for each application area corresponding to category information input from the user among the word dictionary 21 for each application area. According to this, it is possible to significantly shorten the dictionary search time compared to the case of using the existing common library (single library). In addition, the processing of adjectives and verbs is also limited to words limited by dictionaries, so efficiency can be improved.

The morpheme analysis unit 20 is further equipped with a word dictionary learner 23 for deriving words included in the word dictionary 21 for each application area through machine learning for the training data set. Can be.

The analysis unit 30 extracts and outputs analysis information for a social review using the information derived by the above morphological analysis. As the analysis information for the social review, related keywords, related images, popularity indexes, and recognition information may be mentioned.

According to the above morphological analysis method, it is possible to shorten the time to process 500 reviews within 1 minute to 5 seconds, thereby enabling real-time analysis for each place.

Meanwhile, the social analysis system according to the present embodiment may further include a noise filter unit 40 for extracting valid data among collected social reviews.

For example, if the search keyword is given as “Tokyo Beauty”, the search results include noises such as “Tokyo Fine Food Right Side Dining House” and “Tokyo Beauty Building 2nd Floor Pediatrics”, so the filtering process to filter them out This is necessary, and the noise filter unit 40 performs such noise filtering.

The noise filter unit 40 is connected between the morpheme analysis unit 20 and the analysis unit 30, and after filtering the data received from the morpheme analysis unit 20, it is transferred to the analysis unit 30. The valid data extracted by the filtering of the noise filter unit 40 may be stored in a valid data database 45 provided separately.

The existing method of extracting valid data considers that the collected social reviews are less related to a given search keyword, and then extracts only reviews that include predefined positive expressions (eg, 'delicious', 'delicious', etc.) As, it can be referred to as so-called negative (Pessimistic) noise filtering. In the case of such negative noise filtering, it is selected as valid data (associated review) only when the review title has “Tokyo Beauty” or a predefined positive expression such as “Life Taste” or “Gangchu”. However, the ratio adopted as valid data is very low, about 5 to 10% of the total.

For example, in place recommendation, if "Place A" with 50 reviews is ranked 1st and "Place B" with 41 reviews is ranked 2nd, "Place B" hires a viral marketer and only reviews 10 If you increase it, the ranking will change. As described above, when the number of reviews serving as a population of statistics is small, not only the reliability of the calculation result is lowered, but also a problem that is vulnerable to viral marketing occurs.

According to the present embodiment, in order to solve such a problem, the noise filter unit 40 considers all collected social reviews as relevant, and then excludes the reviews including the low-related noise patterns to extract valid data. To extract, we use a so-called positive noise filtering method.

These noise patterns include adverbs indicating adjacent locations of places such as "right" and "second floor of the building", simple listing of words such as bus lines and subway lines, and regional franchises such as "Jongno point" and "Gangnam station point". And words to say.

The noise pattern may also be configured differently for each category, and in this case, the noise filter unit 40 may perform filtering by searching the pattern dictionary 41 for each application area including different noise patterns for each category. For example, words such as “sale” and “premium” are treated as key keywords in the real estate application area, but may be recognized as noise patterns in analysis of application areas such as restaurants and tourist attractions. Derivation of noise patterns for each application area can also be performed through machine learning.

According to the present invention, the word dictionary 21 and the pattern dictionary 41 provided separately for each category may be configured to be modularized for each category. For example, the first module in which the word dictionary 21 and the pattern dictionary 41 corresponding to a restaurant are modularized, and the application corresponding to the tourist attraction is modularized, and the word dictionary 21 and the pattern dictionary corresponding to the tourist attraction A second module in which (41) is modularized may be separately provided. Here, when adding and applying the real estate application area, the word dictionary 21 and the pattern dictionary 41 corresponding to the real estate only need to add a modularized third module, so that the application area and application field can be easily expanded. .

Meanwhile, the social review analysis system according to the present embodiment may further include a translation unit 50 that translates analysis information about social reviews into other languages and outputs translation information. In this way, the multi-language support for related information such as related keywords and recognition information can be used to expand the level of analysis information to foreigners.

The translation unit 50 of the present embodiment basically uses an online translator (eg, Google Translate) for multilingual support, and the translation error problem (eg, “Osam Bulgogi”) of the online translator is “5-3. bulgogi ”, translated“ sex support restaurant ”into“ sex support restaurant ”), a separate proper noun dictionary 51 is provided.

The translation unit 50 includes a proper noun dictionary 51 including translation information about a proper noun, and a translator 52 that searches for the proper noun dictionary 51 and performs translation on the proper noun. The proper noun dictionary 51 may be provided for each of a plurality of languages, and the translator 52 uses a proper noun dictionary for the proper noun part (eg, “Osam Bulgogi”, “Sung Ji-won”, etc.) of the analysis information. Translation is performed by using the online translator 200 through the translator API 53 for the remaining part except the proper noun.

2 is a flowchart illustrating a method for analyzing social reviews of a place according to an embodiment of the present invention.

Referring to FIG. 2, the social review analysis method according to this embodiment will be described in detail. First, the social review collection unit 10 receives a search keyword for a place from a user (S10). 3 illustrates an example of an input page for receiving search keywords and category information. According to this, the search keyword (mutual name, address, etc.) is input through typing, and the category information (restaurant, cafe, lodging facility, tourist attraction, etc.) is inputted by selecting one of the preset menus. Is illustrated.

Next, the social review collection unit 10 collects social reviews of the search keywords received from the user from the social site 100 (S20).

Next, the morpheme analyzer 22 uses the word dictionary 21 for each application area corresponding to the category information input from the user among the word dictionary 21 for each application area provided for each category so that different words are included for each category. Morphological analysis for the social review is performed (S30).

Next, the noise filter unit 40 considers all the collected social reviews as relevant, and then extracts valid data for outputting analysis information by excluding reviews that include noise patterns with low association (S40, positive) Noise filtering). As described above, the noise filter unit 40 filters noise by searching the pattern dictionary 41 for each application area including different noise patterns for each category.

4 illustrates an example of outputting valid data (associated review) after noise filtering for a social review related to the present invention, according to which the summary information and the original text link for the related review are output. . According to the present invention, it is possible to increase the ratio adopted as a related review in the noise filtering step from 5 to 10% to 30% or more.

Next, the analysis unit 30 extracts and outputs analysis information for the social review using the information derived from the above (S50). The analysis information for the social review may include related keywords, related images, popularity index, and recognition information.

FIG. 5 illustrates an example of an output page outputting analysis information for a social review related to the present invention, and accordingly, as an analysis information, an example of outputting an increase / decrease / trend graph of a related keyword or a popularity index for a search place is output. It is done.

Next, if necessary, the step of translating the analysis information for the social review in another language may be additionally performed (S51). As for the translation of the analysis information, as described above, a separate proper noun dictionary 51 is searched for and the translation of the proper noun is performed, and then the online translator 200 is used to translate the remaining parts except the proper noun. It can be done through the method of performing.

On the other hand, in the social review analysis method of the place described above, at least one of the valid data extraction step (S40) and the analysis information translation step (S60) may be omitted if necessary.

The method and system for analyzing social reviews of the place described above may be implemented in the form of program instructions that can be executed through various computer means and may be recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiments or may be known and usable by 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, DVDs, and magnetic media such as floptical disks. Hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like, are included. Examples of program instructions include high-level language code that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Although described above with reference to specific embodiments of the present invention, those skilled in the art may vary the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. You can understand that it can be modified and changed.

10: social review collection unit 20: morpheme analysis unit
21: Word dictionary by application area 22: Morphological analyzer
23: word dictionary learning unit 30: analysis unit
40: noise filter unit 41: pattern dictionary for each application area
50: translation unit 51: proper noun dictionary
52: translator

Claims (13)

Receiving a search keyword and category information regarding a place from a user through a social review collection unit;
Collecting a social review of the search keyword through the social review collection unit;
Performing a morpheme analysis on the social review through a morpheme analysis unit;
Through the noise filter unit, after all collected social reviews are regarded as relevant, extracting valid data for outputting analysis information by excluding reviews that include a low noise pattern; And
And extracting and outputting analysis information for the social review using the information derived by the morpheme analysis and the extracted valid data through an analysis unit,
A word dictionary for each application area containing different words for each category and a pattern dictionary for each application area containing different noise patterns for each category are provided for each category,
The morpheme analysis unit performs morpheme analysis using a word dictionary for each application area corresponding to category information input from a user among the word dictionary for each application area,
The noise filter unit searches for a pattern dictionary for each application area corresponding to category information received from a user among the pattern dictionary for each application area, and extracts valid data.
According to claim 1,
The category information includes one or more of a restaurant, a cafe, a lodging facility, and a tourist attraction.
According to claim 1,
The analysis information for the social review includes one or more of related keywords, associated images, popularity index, and recognition information.
delete delete According to claim 1,
Through the translation unit, further comprising the step of translating the analysis information for the social review in another language,
The translation step,
Performing a translation for the proper noun by searching for a dictionary of proper nouns provided to include translation information about the proper noun through a translator;
And translating the remaining portion of the place except the proper noun through an online translator.
A social review collection unit that receives search keyword and category information regarding a place from a user and collects social reviews for the search keyword;
A morpheme analysis unit that performs a morpheme analysis for the social review;
A noise filter unit that extracts valid data for outputting analysis information by excluding reviews that include a noise pattern having a low association after considering all collected social reviews as relevant; And
And an analysis unit for extracting and outputting analysis information for the social review using the information derived by the morpheme analysis and the extracted valid data,
The morpheme analysis unit,
A dictionary of words for each application area provided for each category and including different words for each category; And
And a morpheme analyzer for performing a morpheme analysis on the social review by using a word dictionary for each application region corresponding to category information input from a user among the word dictionary for each of the plurality of application regions,
The noise filter unit searches for a pattern dictionary for each application area corresponding to category information input from a user among pattern dictionary for each application area including different noise patterns for each category, and performs filtering on the place's social review. system.
The method of claim 7, wherein the morpheme analysis unit,
The social review analysis system of the place further comprising a word dictionary learner to derive words included in the word dictionary for each application area through machine learning on test data.
delete delete The method of claim 7,
The word dictionary and pattern dictionary for each application area are configured to be modularized for each category.
The method of claim 7,
Further comprising a translation unit for translating the analysis information for the social review in another language,
The translation unit,
A proper noun dictionary containing translation information about the proper noun; And
The social noun analysis of the place, wherein the proper noun part of the analysis information includes a translator that performs translation by using the proper noun dictionary, and the translation is performed through an online translator for the remaining parts except the proper noun. system.
The method of claim 12,
The proper noun dictionary is a social review analysis system of a place, characterized in that provided for each of a plurality of languages.

Images