KR102299618B1 - Apparatus and method for matching review advertisement - Google Patents

Abstract

Disclosed are a review advertisement matching apparatus and method, and a recording medium capable of matching a reviewer that meets the requirements of an advertiser based on the entity name of the review text. The review advertisement matching method according to an embodiment of the present invention extracts a set of entity names from review texts written by reviewers by the classification attribute vector extraction module, and based on the entity name classification dictionary tagged with the classification attribute of the entity name extracting a reviewer classification attribute vector by classifying the types of the entity names; and by a matching module, extracting a reviewer attribute vector by synthesizing the reviewer classification attribute vector, obtaining an advertiser attribute vector by analyzing the requirements of an advertiser, and an advertiser based on the similarity between the reviewer attribute vector and the advertiser attribute vector and matching reviewers.

Description

Apparatus and method for review advertisement matching {APPARATUS AND METHOD FOR MATCHING REVIEW ADVERTISEMENT}

The present invention relates to a review advertisement matching apparatus and method, and more particularly, to a review advertisement matching apparatus and method capable of matching a reviewer who meets the requirements of an advertiser based on the entity name of the review text written by the reviewer .

With the activation of online shopping malls, social network services (SNS), blogs, and the like, the importance of review advertisements by reviewers is increasing. However, the recent review advertisement market has reached an oversupply state due to the accessibility of blogs and SNS services, and advertisers have difficulty in selecting an advertisement provider (reviewer) capable of producing high-quality review advertisements.

Due to overheating competition in the review advertising market, reviewers are trying to stand out by relying on quantitative indicators for review text (number of daily visitors, number of subscribers, number of likes, etc.), but selection of reviewers based on these quantitative indicators It causes resource consumption to simply increase quantitative indicators rather than the quality of the advertisement itself, which is a factor hindering the qualitative growth of the overall review advertisement market.

An object of the present invention is to provide a review advertisement matching apparatus and method, and a recording medium capable of matching reviewers who meet the requirements of advertisers based on the entity name of the review text written by the reviewers.

The review advertisement matching method according to an embodiment of the present invention extracts a set of entity names from review texts written by reviewers by the classification attribute vector extraction module, and based on the entity name classification dictionary tagged with the classification attribute of the entity name extracting a reviewer classification attribute vector by classifying the types of the entity names; and by a matching module, extracting a reviewer attribute vector by synthesizing the reviewer classification attribute vector, obtaining an advertiser attribute vector by analyzing the requirements of an advertiser, and an advertiser based on the similarity between the reviewer attribute vector and the advertiser attribute vector and matching reviewers.

The matching of the advertiser and the reviewer may include: extracting the reviewer attribute vector from the reviewer classification attribute vector by an attribute vector extraction module; obtaining, by the attribute vector extraction module, the advertiser attribute vector by analyzing the advertiser's requirements; calculating, by a vector similarity analysis module, a similarity between the reviewer attribute vector and the advertiser attribute vector; and recommending, by the vector similarity analysis module, a reviewer matching the advertiser's demand based on the similarity.

Calculating the similarity may include calculating a matching rate between the advertiser and the reviewers. The step of recommending the reviewers may include recommending the reviewers to the advertiser in the order of the highest matching rate.

A review advertisement matching method according to an embodiment of the present invention comprises the steps of defining attributes of each of the entity names by an attribute definition unit; and, by the review quality evaluator, based on the ratio of the attributes included in the review text among the attributes and the frequency of new attributes in the review text that are not included in the attributes defined in the attribute definition unit, the reviewer's The method may further include evaluating the review quality and providing the reviewer's review quality to the advertiser.

The attributes may include at least two or more of performance, design, delivery quality, and cost related to the entity name.

The review advertisement matching method according to an embodiment of the present invention may further include updating, by an attribute updater, the attributes of the entity names by adding the new attribute that is not defined to the entity names. The updating of the attributes may include: extracting attribute words representing attributes of the entity names from the review text for each entity name; extracting new attribute words whose similarity with attributes predefined in the attribute definition unit is lower than a reference value from among the attribute words; clustering the new attribute words by FastText word clustering; and updating the attributes for each entity name by adding an attribute name in a cluster unit for each entity name.

The review advertisement matching method according to an embodiment of the present invention may further include, by the attribute updater, classifying at least one of the attributes of the entity names into sub-attributes.

The step of classifying the sub-attributes may include: determining whether the review text includes both a positive sentiment evaluation and a negative sentiment evaluation for the review target attribute of the entity name; extracting a first word related to the positive sentiment evaluation and a second word related to the negative sentiment evaluation from the review text, respectively; and classifying the review target attribute into a first sub-attribute related to the positive sentiment evaluation and a second sub-attribute related to the negative sentiment evaluation based on the first word and the second word.

According to an embodiment of the present invention, there is provided a computer-readable recording medium in which a program for executing the review advertisement matching method is recorded.

The review advertisement matching device according to an embodiment of the present invention extracts a set of entity names from review texts written by reviewers, and classifies the types of entity names based on the entity name classification dictionary in which the classification attribute of entity names is tagged. a classification attribute vector extraction module for extracting a reviewer classification attribute vector; and extracting a reviewer attribute vector by synthesizing the reviewer classification attribute vector, obtaining an advertiser attribute vector by analyzing the requirements of the advertiser, and matching advertisers and reviewers based on the similarity between the reviewer attribute vector and the advertiser attribute vector Includes matching module.

The matching module may include: an attribute vector extraction module for extracting the reviewer attribute vector from the reviewer classification attribute vector, and analyzing the advertiser's requirements to obtain the advertiser attribute vector; and a vector similarity analysis module that calculates a similarity between the reviewer attribute vector and the advertiser attribute vector, and recommends a reviewer matching the advertiser's demand based on the similarity.

The vector similarity analysis module calculates a matching rate between the advertiser and the reviewers; And it may be configured to recommend a reviewer to the advertiser in the order of the highest matching rate.

A review advertisement matching apparatus according to an embodiment of the present invention includes: an attribute defining unit defining attributes of each of the entity names; and evaluates the review quality of the reviewer based on the ratio of the attributes included in the review text among the attributes and the frequency of new attributes in the review text that are not included in the attributes defined in the attribute definition unit, It may further include a review quality evaluation unit for providing the review quality of the advertiser.

The review advertisement matching apparatus according to an embodiment of the present invention may further include an attribute update unit for updating the attributes of the entity names by adding the new attribute that is not defined to the entity names. The attribute update unit extracts, for each entity name, attribute words representing attributes of the entity names from the review text; extracting new attribute words whose similarity with attributes predefined in the attribute definition unit is lower than a reference value among the attribute words; clustering the new attribute words by FastText word clustering; In addition, it may be configured to update the attributes for each entity name by adding an attribute name in a group unit for each entity name.

The review advertisement matching apparatus according to an embodiment of the present invention may further include an attribute updater for classifying at least one attribute among attributes of the entity names into sub-attributes. The attribute update unit determines whether the review text includes both a positive emotional evaluation and a negative emotional evaluation for the review target attribute of the entity name; extracting a first word related to the positive sentiment evaluation and a second word related to the negative sentiment evaluation from the review text, respectively; And based on the first word and the second word, it may be configured to classify the review target attribute into a first sub-attribute related to the positive sentiment evaluation and a second sub-attribute related to the negative sentiment evaluation.

According to an embodiment of the present invention, there is provided a review advertisement matching apparatus and method, and a recording medium capable of matching a reviewer that meets the requirements of an advertiser based on the entity name of the review text written by the reviewer.

1 is a block diagram of a review advertisement matching apparatus according to an embodiment of the present invention.
2 is a detailed configuration diagram of a review advertisement matching apparatus according to an embodiment of the present invention.
3 is a conceptual diagram of an entity name recognition module constituting a review advertisement matching apparatus according to an embodiment of the present invention.
4 is a flowchart of a review advertisement matching method according to an embodiment of the present invention.
5 is a block diagram of a review advertisement matching apparatus according to another embodiment of the present invention.
6 is a flowchart of a review advertisement matching method according to another embodiment of the present invention.
7 is a flowchart of a review advertisement matching method according to another embodiment of the present invention.
8 is a flowchart of a review advertisement matching method according to another embodiment of the present invention.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the art to which the present invention pertains It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

In the present specification, when a part "includes" a certain component, it means that other components may be further included rather than excluding other components unless otherwise stated. As used herein, '~ unit' is a unit that processes at least one function or operation, and may refer to, for example, software, FPGA, or hardware component. A function provided by '~ unit' may be performed separately by a plurality of components, or may be integrated with other additional components. The term '~' in this specification is not necessarily limited to software or hardware, and may be configured to reside in an addressable storage medium, or may be configured to reproduce one or more processors.

1 is a block diagram of a review advertisement matching apparatus according to an embodiment of the present invention. Referring to FIG. 1 , the review advertisement matching apparatus 100 according to an embodiment of the present invention extracts a set of individual names from review texts written by reviewers, extracts a reviewer classification attribute vector, and extracts a reviewer classification attribute vector. Based on the similarity between the reviewer attribute vector and the advertiser attribute vector related to the advertiser's requirements, the matching rate between the advertiser and each reviewer is calculated to match the advertiser and the reviewer. Named entity refers to a word or phrase having a specific meaning in a sentence or document, such as a product name, service name, person, organization, or place name.

2 is a detailed configuration diagram of a review advertisement matching apparatus according to an embodiment of the present invention. 3 is a conceptual diagram of an entity name recognition module constituting a review advertisement matching apparatus according to an embodiment of the present invention. 1 to 3 , the review advertisement matching apparatus 100 according to an embodiment of the present invention may include a classification attribute vector extraction module 120 and a matching module 140 .

In order to classify the category of the reviewer from the review text that the reviewer has written in the past, the classification attribute vector extraction module 120 is configured from various review texts related to individual names such as IT devices, cosmetics, and household items written by the reviewers. The entity name recognition module 122 for extracting a set of entity names may be included. Review texts may be collected by, for example, crawling by a data collection unit (not shown), and may be collected by parsing unstructured data for each pattern of SNS/shopping mall/self-owned reviews.

The collected review texts may be pre-processed by a pre-processing unit (not shown), such as morphological analysis and pre-processing (removing stopwords, extracting roots, etc.). By the preprocessor, meaningless word tokens are removed from review texts to remove stopwords, only meaningful word tokens can be selected, and a root that is a basic form of a meaningful word can be extracted.

The classification attribute vector extraction module 120 may extract the reviewer classification attribute vector by classifying the types of entity names based on the entity name classification dictionary 124 tagged with the classification attribute of the entity name in a domain specialized for review advertisements. The entity name classification dictionary 124 may include entity name classification information such as IT devices, cosmetics, and daily necessities. The classification attribute vector extraction module 120 may generate the reviewer-classification attribute vector by synthesizing the review text-classification attribute vector extracted from the review text written by the reviewer.

The matching module 140 extracts the reviewer attribute vector from the reviewer classification attribute vector, and the attribute vector extraction module 142 for obtaining the advertiser attribute vector by analyzing the requirements of the advertiser, and the similarity between the reviewer attribute vector and the advertiser attribute vector It may include a vector similarity analysis module 144 that matches advertisers and reviewers based on the calculated and calculated similarity, and recommends reviewers matching the advertiser's demand.

The matching module 140 constructs a matrix through the advertiser-classification attribute vector and reviewer-classification attribute vector set generated by the attribute vector extraction module 142, and then performs matching analysis between the advertiser and the reviewer using a similarity algorithm. can do. The vector similarity analysis module 144 outputs the similarity (matching rate) of reviewers to the advertiser as an index such as ranking and similarity rate, and may recommend reviewers to the advertiser in the order of the highest matching rate with the advertiser.

4 is a flowchart of a review advertisement matching method according to an embodiment of the present invention. 1 to 4, in the review advertisement matching method, first, by the entity name recognition module 122, a set of entity names is extracted from various review texts related to IT devices, cosmetics, household goods, etc. written by reviewers, respectively. can be (S10). A set of entity names may be calculated for each reviewer by the entity name recognition module 122 .

When a set of object names is extracted for each reviewer, the classification property vector extraction module 120 classifies the types of object names based on the object name classification dictionary 124 tagged with the classification attribute of the object name, and classifies the reviewers for each reviewer. The attribute vector may be extracted (S20). Next, the reviewer attribute vector may be extracted from the reviewer classification attribute vector by the attribute vector extraction module 142 (S30), and the advertiser's attribute vector may be obtained by analyzing the advertiser's requirements (S40).

The advertiser's classification attribute can be extracted by analyzing the questionnaire data collected from the advertiser and obtaining the advertiser-classification vector. When the reviewer attribute vector and the advertiser attribute vector are obtained, the similarity between the reviewer attribute vector and the advertiser attribute vector is calculated by the vector similarity analysis module 144, and the advertiser and the reviewer are matched based on the similarity between the reviewer attribute vector and the advertiser attribute vector It can be done (S50).

The similarity between the reviewer attribute vector and the advertiser attribute vector may be calculated based on, for example, Euclidean Distance, Manhattan Distance, Minkowski Distance, Cosine Similarity, etc. However, it is not limited thereto.

The vector similarity analysis module 144 may calculate a matching rate between advertisers and reviewers based on the similarity between the reviewer attribute vector and the advertiser attribute vector, and recommend reviewers to advertisers in the order of the highest matching rate with the advertiser (S60). The vector similarity analysis module 144 may provide the reviewer with advertiser information in the order of the highest matching rate among the plurality of advertisers to the reviewer.

Advertisers can receive a recommendation from a reviewer that meets their requirements, and request an advertisement from the reviewer. The reviewer may take measures, such as increasing the frequency of review advertisements related to the advertiser, so as to increase the matching rate with the advertiser by checking the advertiser information with a high matching rate with the reviewer.

According to an embodiment of the present invention, by analyzing the review advertisements that the reviewer has published in the past, the reviewers can classify which category of advertisements they are suitable for, and through the classified results, the reviewers suitable for the needs of the advertisers are matched to provide high-quality reviews. By producing advertisements, you can ensure successful review advertisements.

In addition, according to an embodiment of the present invention, a review between an advertiser and a reviewer by analyzing several attributes included in one review text by classifying text through entity name recognition compared to the existing system for classifying text through keyword extraction Ad matching may be performed.

5 is a block diagram of a review advertisement matching apparatus according to another embodiment of the present invention. 6 is a flowchart of a review advertisement matching method according to another embodiment of the present invention. 5 and 6 , the review advertisement matching apparatus 100 according to an embodiment of the present invention further includes an attribute definition unit 160 , a review quality evaluation unit 180 , and an attribute update unit 200 . There is a difference from the embodiment described above.

The attribute definition unit 160 may define categories of various entity names (review subjects such as product/service to be reviewed) and attributes of each entity name (S70). The attributes are unique attributes of each entity name such as a product or service subject to review, and may include at least two or more of performance, design, delivery quality, and cost related to the entity name such as a product or service subject to review.

When the product to be reviewed is a smartphone, the attributes of the smartphone mean various characteristics of the object, such as performance, screen, and weight of the smartphone, for example. The attributes of the entity name may be defined by the attribute definition unit 160 extracting attribute information related to the entity name from the learning data.

In an embodiment, the properties of the review object may be defined by clustering them by word clustering based on FastText, a representative property name is given to each cluster, and adjacent words may be defined in the order in which the representative property name and the vector distance are close to each other. have. It is also possible for the administrator to directly assign or modify the attributes of each entity name by using the domain knowledge of the administrator who manages the review advertisement matching device 100 .

For review analysis and review advertisement matching in units of attributes, the review advertisement matching apparatus 100 uses review data or learning data other than the review data to perform n-gram-based FastText word clustering. A word cluster can be generated by vectorizing and clustering words by word embedding.

In order for the computer to understand the natural language and process it efficiently, it is necessary to properly transform the natural language so that the computer can understand it. To do this, vectorization of words is required, and the representative methods are 'Word2Vec' and 'FastText'. FastText is a library for learning word embedding and text classification, and it is one of the methods to make words into vectors. Unlike 'Word2vec', which considers a word as an indivisible unit, FastText considers multiple words to exist in one word, and performs learning in consideration of multiple subwords in a word.

In 'FastText', each word can be represented by an n-gram, which means a sequence of n consecutive words. How many words are separated can be determined according to how many n is determined. For example, in the case of a tri-gram corresponding to n = 3, each word may be divided into three internal words and embedded. Unlike 'Word2Vec', 'FastText' can calculate similarity with other words even for subwords and unknown words (Out of Vocabulary). For example, when the word 'battery charge' is not learned, if you have learned 'battery' and 'charge' with n-grams of other words, you can create an embedding vector of 'battery charge' .

A cluster of similar words can be created by performing clustering based on a vector distance such as the Euclidean distance of words vectorized through n-grams. For example, attributes such as 'delivery', 'courier', 'design', and 'performance' may be vectorized as follows, and words having close vector distances may be clustered into adjacent words.

The review quality evaluation unit 180 determines the reviewer's rating based on the ratio of the attributes included in the review text among the defined attributes and the frequency of new attributes in the review text not included in the attributes defined in the attribute definition unit 160 . The review quality may be evaluated (S80), and the reviewer's review quality may be provided to the advertiser (S90).

For example, the more various attributes (eg, performance, design, delivery, cost, etc.) related to the entity name (product/service, etc.) related to the advertiser are evaluated in the review text written by the reviewer, the reviewer's review quality This can be highly appreciated. Also, even when the reviewer evaluates new properties other than the defined properties, the review quality of the reviewer may be highly evaluated.

According to an embodiment of the present invention, the reviewer analyzes the attributes analyzed in relation to the entity name in the review text (the frequency of attributes evaluated in the review text, the number of new attributes, etc.), evaluates the review quality of the reviewer, and sends it to the advertiser By providing it, the advertiser can help determine a reviewer by considering the reviewer's review quality.

7 is a flowchart of a review advertisement matching method according to another embodiment of the present invention. 5 and 7 , the attribute update unit 200 may update the attributes of entity names by adding new undefined attributes to entity names through steps S110 to S140.

The attribute updater 200 first extracts attribute words representing attributes of entity names from the review text for each entity name (S110), and among the extracted attribute words, similarity with attributes predefined in the attribute definition unit 160 New attribute words lower than the reference value may be extracted (S120).

The attribute update unit 200, for example, when the subject name of the object to be reviewed is a smartphone, the attribute word related to 'folding wrinkles' is extracted from the review text, but when the 'folding wrinkles' attribute is not defined with respect to the smartphone , it is possible to update the attributes defined in the smartphone by adding the 'folding fold' attribute for the smartphone.

The attribute update unit 200 clusters new attribute words by n-gram-based FastText word clustering (S130), and adds attribute names in cluster units to individual names for each entity name. The attributes may be updated (S140). The similarity between the attributes defined in the attribute definition unit 160 and attribute words extracted from the review text may be calculated based on a vector distance such as, for example, a Euclidean distance.

According to the embodiment shown in FIG. 7 , it is possible to evaluate the review quality of a reviewer for each of various attributes of an entity name through attribute update that adds a new attribute of a subject to be reviewed (subject name) based on the review text. Therefore, when a new type of entity name (subject name) appears, such as a foldable smartphone, by updating the properties related to the entity name, the review quality of the reviewer is evaluated by various attributes for the entity name, and review advertisement between the reviewer and the advertiser. can match.

8 is a flowchart of a review advertisement matching method according to another embodiment of the present invention. 5 and 8, in the review advertisement matching method according to another embodiment of the present invention, the attribute updater 200 subordinates at least one attribute among attributes of entity names through steps S150 to S170. properties can be classified.

The attribute update unit 200 determines whether the review text includes both the positive emotional evaluation and the negative emotional evaluation for the review target attribute of the entity name (S150), and positive emotional evaluation and When all of the negative emotional evaluations are included, the first word related to the positive emotional evaluation and the second word related to the negative emotional evaluation may be extracted from the review text (S160).

The attribute update unit 200 sets the review target attribute to the first sub-attribute and negative emotion related to the positive emotional evaluation based on the first word extracted in relation to the positive emotional evaluation and the second word extracted in relation to the negative emotional evaluation. It can be classified into a second sub-attribute related to evaluation (S170).

For example, with respect to the design attributes of a car, words such as 'inside' and 'interior' are extracted in relation to positive emotional evaluation, and words such as 'external' and 'exterior' are extracted in relation to negative emotional evaluation. In this case, the design properties of a vehicle may be classified into sub-properties of 'exterior design' and 'interior design'.

According to the embodiment shown in FIG. 8 , the reviewer's review quality by various attributes and sub-attributes for the review subject through attribute update that subdivides and classifies the attributes of the review object into sub-attributes based on the review text written by the reviewer can be analyzed to match advertisements between advertisers and reviewers.

According to an embodiment of the present invention, by evaluating the matching rate with the advertiser based on the attribute word (subject name) from the review data, and analyzing the reviewer's review quality, the Reviewers can be recommended, and information about advertisers with a high matching rate can be provided to reviewers to help reviewers select a target for review advertisements.

The embodiments described above may be implemented by a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the apparatus, method, and component described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). Array), a Programmable Logic Unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose or special purpose computers.

The processing device may run an operating system and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, the processing device is sometimes described as being used, but one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It will be understood that this may include

For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a Parallel Processor. The software may include a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device.

The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software.

Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CDROMs and DVDs, and ROM, RAM, and flash memory. Hardware devices specially configured to store and execute program instructions, such as, etc. are included. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible from the above description by those skilled in the art. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result. Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

100: review ad matching device
120: Classification attribute vector extraction module
122: entity name recognition module
124: Entity Name Classification Dictionary
140: matching module
142: attribute vector extraction module
144: vector similarity analysis module

Claims (15)

By the classification property vector extraction module, a set of object names is extracted from the review text written by each reviewer, and the type of object names are classified based on the object name classification dictionary tagged with the classification property of the object name to obtain a reviewer classification property vector. extracting; and
By the matching module, the reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the advertiser based on the similarity between the reviewer attribute vector and the advertiser attribute vector matching reviewers;
defining attributes of each of the entity names by an attribute definition unit; and
The reviewer's review by the review quality evaluator based on the ratio of the attributes included in the review text among the attributes and the frequency of new attributes in the review text that are not included in the attributes defined in the attribute definition unit The review advertisement matching method further comprising the step of evaluating the quality and providing the reviewer's review quality to the advertiser. According to claim 1,
The step of matching the advertiser and the reviewer is,
extracting, by an attribute vector extraction module, the reviewer attribute vector from the reviewer classification attribute vector;
obtaining, by the attribute vector extraction module, the advertiser attribute vector by analyzing the advertiser's requirements;
calculating, by a vector similarity analysis module, a similarity between the reviewer attribute vector and the advertiser attribute vector; and
and recommending, by the vector similarity analysis module, a reviewer matching the advertiser's demand based on the similarity. 3. The method of claim 2,
Calculating the similarity includes calculating a matching rate between the advertiser and the reviewers,
The step of recommending the reviewer is a review advertisement matching method comprising the step of recommending a reviewer to the advertiser in the order of the highest matching rate. delete The method of claim 1 , wherein the attributes include at least two or more of performance, design, delivery quality, and cost related to the entity name. By the classification property vector extraction module, a set of object names is extracted from the review text written by each reviewer, and the type of object names are classified based on the object name classification dictionary tagged with the classification property of the object name to obtain a reviewer classification property vector. extracting; and
By the matching module, the reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the advertiser based on the similarity between the reviewer attribute vector and the advertiser attribute vector matching reviewers;
defining attributes of each of the entity names by an attribute definition unit; and
The method further comprising the step of updating the attributes of the entity names by adding a new attribute that is not defined to the entity names by the attribute updating unit,
Updating the properties includes:
extracting attribute words representing attributes of the entity names from the review text for each entity name;
extracting new attribute words whose similarity with attributes predefined in the attribute definition unit is lower than a reference value from among the attribute words;
clustering the new attribute words by FastText word clustering; and
and updating the attributes for each entity name by adding an attribute name in a cluster unit for each entity name. By the classification property vector extraction module, a set of object names is extracted from the review text written by each reviewer, and the type of object names are classified based on the object name classification dictionary tagged with the classification property of the object name to obtain a reviewer classification property vector. extracting;
By the matching module, the reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the advertiser based on the similarity between the reviewer attribute vector and the advertiser attribute vector matching reviewers; and
classifying, by an attribute updater, at least one attribute among attributes of the entity names into sub-attributes,
The step of classifying into the sub-properties is,
determining whether the review text includes both a positive emotional evaluation and a negative emotional evaluation for the review target attribute of the entity name;
extracting a first word related to the positive sentiment evaluation and a second word related to the negative sentiment evaluation from the review text, respectively; and
based on the first word and the second word, classifying the review target attribute into a first sub-attribute related to the positive sentiment evaluation and a second sub-attribute related to the negative sentiment evaluation; . A computer-readable recording medium in which a program for executing the review advertisement matching method of any one of claims 1 to 3, 5 to 7 is recorded. a classification attribute vector extraction module for extracting a set of entity names from review texts written by reviewers, classifying the types of entity names based on an entity name classification dictionary tagged with classification attributes of entity names, and extracting a reviewer classification attribute vector; and
The reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the reviewer are matched based on the similarity between the reviewer attribute vector and the advertiser attribute vector. contains a module;
an attribute definition unit defining attributes of each of the entity names; and
Evaluating the review quality of the reviewer based on the ratio of the attributes included in the review text among the attributes and the frequency of new attributes in the review text that are not included in the attributes defined in the attribute definition unit, Review advertisement matching device further comprising a review quality evaluation unit for providing the review quality to the advertiser. 10. The method of claim 9,
The matching module is
an attribute vector extraction module for extracting the reviewer attribute vector from the reviewer classification attribute vector, and analyzing the advertiser's requirements to obtain the advertiser attribute vector; and
and a vector similarity analysis module for calculating a similarity between the reviewer attribute vector and the advertiser attribute vector, and recommending a reviewer matching the advertiser's demand based on the similarity. 11. The method of claim 10,
The vector similarity analysis module,
calculating a matching rate between the advertiser and the reviewers; and
A review advertisement matching device configured to recommend reviewers to the advertiser in the order of the highest matching rate. delete The apparatus of claim 9 , wherein the attributes include at least two or more of performance, design, delivery quality, and cost related to the entity name. a classification attribute vector extraction module for extracting a set of entity names from review texts written by reviewers, classifying the types of entity names based on an entity name classification dictionary tagged with classification attributes of entity names, and extracting a reviewer classification attribute vector;
The reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the reviewer are matched based on the similarity between the reviewer attribute vector and the advertiser attribute vector. module;
an attribute definition unit defining attributes of each of the entity names; and
and an attribute update unit that updates the attributes of the entity names by adding new attributes that are not defined to the entity names;
The attribute update unit,
extracting attribute words representing attributes of the entity names from the review text for each entity name;
extracting new attribute words whose similarity with attributes predefined in the attribute definition unit is lower than a reference value among the attribute words;
clustering the new attribute words by FastText word clustering; and
A review advertisement matching device configured to update the attributes for each entity name by adding an attribute name in a group unit for each entity name. a classification attribute vector extraction module for extracting a set of entity names from review texts written by reviewers, classifying the types of entity names based on an entity name classification dictionary tagged with classification attributes of entity names, and extracting a reviewer classification attribute vector;
The reviewer attribute vector is extracted by synthesizing the reviewer classification attribute vector, the advertiser attribute vector is obtained by analyzing the requirements of the advertiser, and the advertiser and the reviewer are matched based on the similarity between the reviewer attribute vector and the advertiser attribute vector. module; and
and an attribute updater for classifying at least one attribute among attributes of the entity names into sub-attributes,
The attribute update unit,
determining whether the review text includes both a positive emotional evaluation and a negative emotional evaluation for the review target attribute of the entity name;
extracting a first word related to the positive sentiment evaluation and a second word related to the negative sentiment evaluation from the review text, respectively; and
A review advertisement matching device configured to classify the review target attribute into a first sub-attribute related to the positive sentiment evaluation and a second sub-attribute related to the negative sentiment evaluation based on the first word and the second word.

Images