KR20230154710A - Method, computer device, and computer program to provide search result using blocks of search intent unit - Google Patents

Abstract

A method, computer device, and computer program for providing search results using blocks in search intent units are disclosed. A method of providing search results includes generating a block in search intent units by combining a topic and a genre; and providing personalized search results for a search term given by the user using the block selected based on context related to the user.

Description

Method, computer device, and computer program for providing search results using blocks of search intent units {METHOD, COMPUTER DEVICE, AND COMPUTER PROGRAM TO PROVIDE SEARCH RESULT USING BLOCKS OF SEARCH INTENT UNIT}

The description below relates to technology that provides personalized search results.

A search engine is software or a website that helps you easily find information on the Internet. These search engines search databases through search terms entered by users, extract search results, and provide them to users.

As an example of a technology that provides such search results, Korean Patent Publication No. 10-2004-0079604 (published on September 16, 2004) discloses a technology for generating a search result list for a search term in a search engine. .

By using blocks, a new search result unit that subdivides search intents, search results that take into account individual intents and tastes can be provided.

Topics and genres can be extracted through document classification, and blocks in search intent units can be created by combining topics and genres.

For each keyword, you can create a block with a combination of topics and genres corresponding to that keyword, and then match the documents to be exposed through the block to build data that can be used for real-time search.

Search results can be provided in personalized blocks by considering the user's search intent based on the search term and the user's personal preferences according to the context.

Optimized by freely combining the correct answer block containing correct search results according to the search intent, the navigation block containing customized search results based on user taste, and the responsive block containing real-time search results based on user reaction. can provide search results.

A personalized search method executed on a computer device, wherein the computer device includes at least one processor configured to execute computer-readable instructions included in a memory, and wherein the personalized search method includes, by the at least one processor, a topic Creating a search intent unit block using a combination of topic and genre; and providing, by the at least one processor, personalized search results for a search term given by a user using the block selected based on context related to the user.

According to one aspect, the generating step includes extracting the topic and the genre through document analysis of documents on the Internet; determining a naming of the block, which is a search result unit, by combining the topic and the genre; And for each of the blocks, it may include linking a document to be exposed as a search result through the corresponding block.

According to another aspect, the generating step includes selecting a keyword for block creation; generating the block, which is a search result unit, by combining the topic and the genre for each of the keywords; And for each of the blocks, it may include linking a document to be exposed as a search result through the corresponding block.

According to another aspect, the generating step may include periodically updating the block information for each keyword and the document information for each block in a cache for real-time search.

According to another aspect, the generating step includes extracting the topic for intent classification through topic modeling for documents on the Internet; extracting the genre to subdivide the topic by classifying the document according to document format; determining a naming of the block, which is a search result unit, by combining the topic and the genre; And for each of the blocks, it may include linking a document to be exposed as a search result through the corresponding block.

According to another aspect, the providing step includes: searching for a block corresponding to the search word among the blocks; and determining a ranking of the block corresponding to the search word based on at least one of the user's profile information and history information.

According to another aspect, the searching step may search for a block of names that include a keyword that matches the search term or a keyword that is semantically similar to the search term.

According to another aspect, the search step may search for a block of names containing a keyword corresponding to the search term using a deep matching model based on a neural network that learns sparse expressions for keywords and documents.

According to another aspect, the searching step includes collecting extended keywords related to the block using documents pre-matched to the block; And it may include searching for a block corresponding to the search term using the name of the block and the extended keyword.

According to another aspect, the determining step includes a ranking model using a user feature representing an embedding vector for the user information, a search term feature representing an embedding vector for the search term, and a block feature representing an embedding vector for the block. You can use to perform personalized ranking of blocks corresponding to the search term.

A computer program stored in a computer-readable recording medium is provided to execute the personalized search method on a computer.

A computer device comprising: at least one processor configured to execute computer-readable instructions included in a memory, wherein the at least one processor generates a block of search intent units using a combination of a topic and a genre, and retrieves information from a user. A computer device is provided that provides personalized search results using the block selected based on context related to the user for a given search term.

According to embodiments of the present invention, search results that take into account individual intent and taste can be provided using blocks, a new search result unit that subdivides search intent.

According to embodiments of the present invention, topics and genres can be extracted through document classification, and blocks in units of search intents can be created by combining topics and genres rather than the type or source of content.

According to embodiments of the present invention, a block can be created for each keyword with a combination of topics and genres corresponding to the keyword, and then documents to be exposed through the block can be matched to construct data that can be used for real-time search.

According to embodiments of the present invention, search results can be provided in personalized blocks by considering the user's search intent based on the search word and the user's personal preferences according to the context.

According to embodiments of the present invention, a correct answer block containing correct answer search results according to search intent, a search block containing customized search results based on user taste, and a response containing real-time search results based on user reaction. Type blocks can be freely combined to provide optimized search results.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention.
Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention.
Figure 3 shows the overall system structure for a personalized search service in one embodiment of the present invention.
Figure 4 is a flowchart showing an example of an intent classification method in an embodiment of the present invention.
Figure 5 shows an example of document structuring through intent classification in one embodiment of the present invention.
Figure 6 is a flowchart showing an example of an intent block generation method in an embodiment of the present invention.
7 to 10 are example diagrams for explaining the naming process of an intent block in an embodiment of the present invention.
Figure 11 is a flowchart showing an example of a method for providing search results in one embodiment of the present invention.
Figure 12 is an example diagram for explaining the intent block search process in one embodiment of the present invention.
Figures 13 to 20 show examples of user interface screens that display search results in intent block units according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

Embodiments of the present invention relate to technology for providing personalized search results.

Embodiments including those specifically disclosed in this specification can provide search results that take into account the individual's intent and taste using blocks, which are a new search result unit that subdivides the search intent.

The personalized search system according to embodiments of the present invention may be implemented by at least one computer device, and the personalized search method according to embodiments of the present invention is performed through at least one computer device included in the personalized search system. It can be. At this time, the computer program according to an embodiment of the present invention may be installed and driven in the computer device, and the computer device may perform the personalized search method according to the embodiment of the present invention under the control of the driven computer program. . The above-described computer program can be combined with a computer device and stored in a computer-readable recording medium for executing a personalized search method on a computer.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention. The network environment in FIG. 1 shows an example including a plurality of electronic devices 110, 120, 130, and 140, a plurality of servers 150 and 160, and a network 170. Figure 1 is an example for explaining the invention, and the number of electronic devices or servers is not limited as in Figure 1. In addition, the network environment in FIG. 1 only explains one example of environments applicable to the present embodiments, and the environment applicable to the present embodiments is not limited to the network environment in FIG. 1.

The plurality of electronic devices 110, 120, 130, and 140 may be fixed terminals or mobile terminals implemented as computer devices. Examples of the plurality of electronic devices 110, 120, 130, and 140 include smart phones, mobile phones, navigation devices, computers, laptops, digital broadcasting terminals, Personal Digital Assistants (PDAs), and Portable Multimedia Players (PMPs). ), tablet PC, etc. For example, in FIG. 1, the shape of a smartphone is shown as an example of the electronic device 110. However, in embodiments of the present invention, the electronic device 110 actually communicates with other devices through the network 170 using a wireless or wired communication method. It may refer to one of various physical computer devices capable of communicating with electronic devices 120, 130, 140 and/or servers 150, 160.

The communication method is not limited, and may include not only a communication method utilizing a communication network that the network 170 may include (for example, a mobile communication network, wired Internet, wireless Internet, and a broadcast network), but also short-range wireless communication between devices. For example, the network 170 may include a personal area network (PAN), a local area network (LAN), a campus area network (CAN), a metropolitan area network (MAN), a wide area network (WAN), and a broadband network (BBN). , may include one or more arbitrary networks such as the Internet. Additionally, the network 170 may include any one or more of network topologies including a bus network, star network, ring network, mesh network, star-bus network, tree or hierarchical network, etc. Not limited.

Each of the servers 150 and 160 is a computer device or a plurality of computers that communicate with a plurality of electronic devices 110, 120, 130, 140 and a network 170 to provide commands, codes, files, content, services, etc. It can be implemented with devices. For example, the server 150 may be a system that provides a service (for example, a search service) to a plurality of electronic devices 110, 120, 130, and 140 connected through the network 170.

Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention. Each of the plurality of electronic devices 110, 120, 130, and 140 described above or each of the servers 150 and 160 may be implemented by the computer device 200 shown in FIG. 2. For example, a personalized search system according to embodiments of the present invention may be implemented by the computer device 200 shown in FIG. 2.

As shown in FIG. 2, this computer device 200 may include a memory 210, a processor 220, a communication interface 230, and an input/output interface 240. The memory 210 is a computer-readable recording medium and may include a non-permanent mass storage device such as random access memory (RAM), read only memory (ROM), and a disk drive. Here, non-perishable large-capacity recording devices such as ROM and disk drives may be included in the computer device 200 as a separate permanent storage device that is distinct from the memory 210. Additionally, an operating system and at least one program code may be stored in the memory 210. These software components may be loaded into the memory 210 from a computer-readable recording medium separate from the memory 210. Such separate computer-readable recording media may include computer-readable recording media such as floppy drives, disks, tapes, DVD/CD-ROM drives, and memory cards. In another embodiment, software components may be loaded into the memory 210 through the communication interface 230 rather than a computer-readable recording medium. For example, software components may be loaded into memory 210 of computer device 200 based on computer programs installed by files received over network 170.

The processor 220 may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input/output operations. Commands may be provided to the processor 220 by the memory 210 or the communication interface 230. For example, processor 220 may be configured to execute received instructions according to program code stored in a recording device such as memory 210.

The communication interface 230 may provide a function for the computer device 200 to communicate with other devices (eg, the storage devices described above) through the network 170. For example, a request, command, data, file, etc. generated by the processor 220 of the computer device 200 according to a program code stored in a recording device such as memory 210 is transmitted to the network ( 170) and can be transmitted to other devices. Conversely, signals, commands, data, files, etc. from other devices may be received by the computer device 200 through the communication interface 230 of the computer device 200 via the network 170. Signals, commands, data, etc. received through the communication interface 230 may be transmitted to the processor 220 or memory 210, and files, etc. may be stored in a storage medium (as described above) that the computer device 200 may further include. It can be stored as a permanent storage device).

The input/output interface 240 may be a means for interfacing with the input/output device 250. For example, input devices may include devices such as a microphone, keyboard, or mouse, and output devices may include devices such as displays and speakers. As another example, the input/output interface 240 may be a means for interfacing with a device that integrates input and output functions, such as a touch screen. The input/output device 250 may be configured as a single device with the computer device 200.

Additionally, in other embodiments, computer device 200 may include fewer or more components than those of FIG. 2 . However, there is no need to clearly show most prior art components. For example, the computer device 200 may be implemented to include at least some of the input/output devices 250 described above, or may further include other components such as a transceiver, a database, etc.

Hereinafter, a specific embodiment of the method and system for providing search results using blocks in search intent units will be described.

Figure 3 shows the overall system structure for a personalized search service in one embodiment of the present invention.

Figure 3 shows an intent classification server 310, an intent block server 320, and a search server 330. The intent classification server 310, the intent block server 320, and the search server 330 may be implemented by the computer device 200 described with reference to FIG. 2 .

For example, the intent classification server 310, the intent block server 320, and the search server 330 may be implemented as separate systems to provide a personalized search service through inter-system connection. Depending on the embodiment, the intent classification server 310, the intent block server 320, and the search server 330 may be implemented as one system, for example, the intent classification server 310, the intent The block server 320 and the search server 330 may be software modules installed and driven on the server 150 to provide a personalized search service.

The intent classification server 310 is responsible for classifying documents on the Internet into intent units consumed by users. The intent classification server 310 can identify the intents of users consuming the document through document analysis.

As an example, intents can be identified using topic and genre. In this embodiment, the criteria for classifying documents by intent are the topic of the document (what the document specifically explains) and the genre (what type of content the document contains or how the document is written). (Is the content being conveyed in type?) can be utilized. In other words, the intent classification server 310 can extract topics and genres through document analysis for documents consumed by users, and structures each document into intent units with a two-dimensional structure consisting of topics and genres. can do.

As another example, intents can be identified using specific conditions (e.g., products with good reviews, products frequently purchased by women in their 30s, etc.) or themes (e.g., camping sites suitable for taking children). there is.

Below, an example using topics and genres is described as an embodiment of the present invention, but it is not limited to this and, of course, various user intents can be reflected.

Additionally, when creating an intent block, in addition to the user's intention, new ranking logic (e.g., most recent, etc.), source or form of the document (e.g., short video, etc.), etc. can be additionally considered. These blocks can be used according to the intention of the user consuming the search service. For example, an intent block consisting of only short videos can be provided to a user who wants to consume short videos.

The intent block server 320 is responsible for blocking documents on the Internet into intent units consumed by users. For example, the intent block server 320 may generate a search intent unit block (hereinafter referred to as an 'intent block') as a new search result unit based on topic and genre. At this time, the intent block server 320 can set an intent block corresponding to the keyword for each pre-selected keyword and then match in advance documents matching the intent block among documents from various sources existing on the Internet. there is. The intent block server 320 can create a block for each keyword with a combination of topics and genres corresponding to the keyword, and then match documents to be exposed through the block to construct data that can be used for real-time search.

The search server 330 is responsible for providing real-time search using pre-created intent blocks. The search server 330 may block and provide search results for a search term by intent unit rather than document source or type unit. When a search term is input from a user, the search server 330 may select an intent block that matches the search term and then combine the selected intent blocks to provide search results. In particular, when combining intent blocks that match the search term, the search server 330 includes profile information including the user's gender, age, region, region, etc., as well as the user's search history, shopping history, and content consumption history. Personalized search results can be provided through ranking using historical information.

Figure 4 is a flowchart showing an example of an intent classification method in an embodiment of the present invention.

The intent classification method according to this embodiment may be performed by the computer device 200 implementing the intent classification server 310. In this case, the processor 220 of the computer device 200 may be implemented to execute control instructions according to the code of an operating system or at least one program included in the memory 210. Here, the processor 220 causes the computer device 200 to perform the steps (S410 to S430) included in the intent classification method of FIG. 4 according to the control command provided by the code stored in the computer device 200. (200) can be controlled.

Referring to FIG. 4, in step S410, the processor 220 may extract a topic, which is one of the criteria for classifying document intents, through document classification targeting documents being consumed by users. For example, the processor 220 may select a topic by analyzing documents belonging to each category for each category corresponding to the document classification system. For example, the processor 220 may select a topic using search words used when searching for a document, keywords that frequently appear in the document, categories or tags set by the user for the document, etc. At this time, the processor 220 can extract topics by selecting main keywords for documents belonging to each category through TF-IDF (Term Frequency-Inverse Document Frequency) or document group clustering. For topic modeling, deep learning-based classifiers such as BERT (Bidirectional Encoder Representations from Transformers) and LDA (Latent Dirichlet Allocation) can be used. As another example, the processor 220 may create a document group based on document-based network information and extract a topic. One document can have various metadata, and for example, documents can be classified based on metadata such as author information, tags, location information, and category information. Documents with the same metadata can be classified based on metadata, under the premise that they will be documents on the same topic. In document classification, you can use a deep learning model that has learned to embed metadata and documents in the same vector space, and you can collect documents with similar embeddings and select topics from the document group. At this time, label noise removal logic can be applied to filter out noise in the data set for learning the deep learning model. For example, you can learn a plurality of classification models (Logistic Regression, Random Forest, MLP, etc.) using different data for labeling documents, and then filter the learning data set using the learned classification models.

In step S420, the processor 220 may extract genre, which is another criterion for classifying document intent, through document classification for documents being consumed by users. The processor 220 can select the genre of the document by classifying the document according to the document format, and select a major genre with high user preference, that is, a large number of documents, among the document genres. This is to further classify documents classified based on topic into more detailed criteria, that is, genre. The processor 220 may classify the document format using various features of the document. Depending on the document format, the features used for classification can be configured differently. For example, the processor 220 can classify a document that compares two or more objects through a classifier that includes text features (title features, body features, etc.) of the document and attribute features using attribute information of the document. there is. For example, attribute information that appears in a document includes attribute frequency values (e.g., number of attribute values, number of sentences that appear, etc.), attribute DF (Document Frequency) or TF (Term Frequency), and attribute category (what category each attribute belongs to). It is possible to create features (i.e., attribute features) to classify the comparative document format using factors such as whether the document belongs to the document type. As another example, the processor 220 may classify a document in a format that recommends at least one object through a classifier that includes text features of the document and features using structure information of the document. The processor 220 can find emphasized styles or parallel content listing structures by analyzing the structure of the document, and can use this document structure information to generate features for classifying recommended document formats. As another example, the processor 220 guides the target through a classifier that includes text features, attribute features, structure features, image features, and author-related features of the document (how to, know-how, tip, etc.). Documents can be classified. You can use the text features of the document to classify whether it is a guide document, and in addition to the text features, you can classify high-quality guide documents using various features such as attribute features, structural features, image features, and author-related features. As another example, the processor 220 may classify a later format document through a classifier that includes text features of the document and experience-related features. Experience expressions that appear in documents (e.g., went there, visited, etc.), subject expressions (e.g., friends, family, me, etc.), and time expressions (e.g., 2 days and 1 night, 3 days, 1 month) etc.), opinion expressions (e.g., likes, recommendations, etc.), etc., can be defined and used as features for document classification. Using the experience-related features of the document, you can classify whether it is a review document, and in the case of a review document, you can classify it into a document with a positive opinion and a document with a negative opinion.

In step S430, the processor 220 may classify the document into a two-dimensional structure of topic and genre. The processor 220 can classify documents by intent by structuring them into topics and genres. As shown in FIG. 5, the processor 220 may define a topic 501 and a genre 502 as criteria for classifying documents by intent. By analyzing documents consumed by users, topics 501 and genres 502 to distinguish document intents can be extracted, and through this, documents can be classified according to topics 501 and genres 502. For example, in the case of a document on 'How to set up a tent', it may correspond to a 'Leisure > Camping' document among topics (501) and a 'How-To' document (guide document format) among genres (502). .

The processor 220 can utilize various deep learning models such as BERT model, LDA model, logistic regression model, random forest model, and MLP (multi-layer perceptron) model for topic-based document classification and genre-based document classification, and a single model In addition, an ensemble model that combines two or more models can be used. For example, the final result can be obtained through voting after combining different models, and GNN (graph neural network), etc. can be used to combine two or more classifier models.

Figure 6 is a flowchart showing an example of an intent block generation method in an embodiment of the present invention.

The intent block creation method according to this embodiment may be performed by the computer device 200 implementing the intent block server 320. In this case, the processor 220 of the computer device 200 may be implemented to execute control instructions according to the code of an operating system or at least one program included in the memory 210. Here, the processor 220 causes the computer device 200 to perform the steps (S610 to S630) included in the intent block generation method of FIG. 6 according to the control command provided by the code stored in the computer device 200. The device 200 can be controlled.

Referring to FIG. 6, in step S610, the processor 220 may select a keyword for creating a block. The processor 220 generates a keyword pool for block creation among keywords used as search words on the Internet. For example, it can analyze the frequency of use of search words and select keywords with a frequency of use above a certain level.

In step S620, the processor 220 may perform naming to create an intent block by combining a topic and a genre for each keyword selected in step S610. First, referring to FIG. 7 , the processor 220 may map at least one topic 720 to each keyword 710. Topics 720 may include steady topics (721) that are constantly repeated and trend topics (722) that appear frequently recently. The processor 220 may map a topic 720 suitable for each keyword 710 based on the relationship between the keyword 710 and the topic 720. Next, referring to FIG. 8, the processor 220 combines the topic 720 mapped to the keyword 710 for each keyword 710 and the subdivided genre 830 of each topic 720 to create a block name candidate ( 840) can be made. The genre 830 may include a genre 831 defined based on a document as well as a genre 832 defined based on a source. At this time, the processor 220 may perform filtering on the block name candidate 840 using the document proportion. Referring to FIG. 9, the processor 220 may exclude some of the block name candidates 840 according to the proportion of documents classified into the topic 720 and the genre 830. For example, if the number of documents classified as 'camping site' and genre as 'review', a review document format, is below the threshold, 'camping site review' created by combining topic 'camping site' and genre 'review' ' may be excluded from intent block creation. The processor 220 may select the final block name candidate 950 through block name filtering using document proportion. Additionally, the processor 220 may combine the keywords 710 and the final block name candidates 950 in different orders and then use a language model to determine a more natural name in terms of language structure. For example, as shown in Figure 10, given the keyword 'bicycle', topic 'Han River', and genre 'review', possible naming candidates by varying the order of combination of keyword, topic, and genre are <Hangang Bicycle Review>, <Review. After creating Hangang Bicycle>, <Hangang Bicycle Review>, etc., the final naming (for example, <Han River Bicycle Review>) with the lowest cross-entropy loss can be selected through the prediction probability using a language model (1000). The processor 220 may generate at least one intent block for each keyword 710 through the above-mentioned naming process.

In the naming process in the intent block creation process, naming in the form of a word combination is explained, but it is not limited to this, and various forms of naming, such as a complete sentence type, are possible. For example, given the keyword 'bicycle', topic 'Han River', and genre 'recommendation', such as <Which course is the most popular as a Han River bike course?>, <When is the best time to ride a bike on the Han River?>, etc. You can create a block name using a question-type sentence. You can create a Q&A (Question & Answer) style intent block by naming blocks using question-type sentences.

Referring again to FIG. 6, in step S630, the processor 220 may link a document to be exposed through the block to each intent block for which the final name has been determined. The processor 220 may match documents to be exposed to the block for each intent block for each keyword. For example, the processor 220 can search for documents from various sources on the Internet using the naming of an intent block, and store at least one document included in the search results in the cache of the corresponding block. The processor 220 can enter the block based on documents belonging to the topic and genre of the intent block.

Accordingly, the processor 220 may determine a block name using a combination of topic and genre for major keywords used on the Internet and then generate an intent block by matching a document in advance to the block whose name has been determined. The processor 220 can prepare documents of intent blocks in advance from various sources so that they can be used for real-time search, and at this time, intent block information for each keyword and document information for each intent block are periodically updated in the cache for real-time search. can do.

Meanwhile, in addition to the method of determining the block naming by combining the topic and genre described above, it is possible to create an intent block using other methods. You can create blocks by combining properties that frequently appear in major object keywords. For example, it is a combination of object keywords and frequently appearing attributes, such as (car model name)+'test drive', (food ingredient name)+'storage method', (book name)+'book review'. Properties can be extracted and created by combining expressions that frequently appear in documents related to specific object keywords. Additionally, it is possible to extract key phrases from a corpus of recently created documents and create blocks by grouping documents that share the same key phrases. For example, you can use a deep learning model to generate sentence vectors of documents that share search queries and user tags, average them, and then combine and attach the corresponding tags to similar documents. These various intent block creation methods can be combined or used complementary to each other as needed.

Figure 11 is a flowchart showing an example of a method for providing search results in one embodiment of the present invention.

The method for providing search results according to this embodiment may be performed by the computer device 200 implementing the search server 330. In this case, the processor 220 of the computer device 200 may be implemented to execute control instructions according to the code of an operating system or at least one program included in the memory 210. Here, the processor 220 causes the computer device 200 to perform the steps (S1110 to S1130) included in the search result providing method of FIG. 11 according to the control command provided by the code stored in the computer device 200. (200) can be controlled.

Referring to FIG. 11 , in step S1110, when a search word is input by the user, the processor 220 may search for an intent block corresponding to the search word among pre-created intent blocks. For example, the processor 220 may search for an intent block whose name includes a keyword that matches the search term using a traditional term matching method. As another example, the processor 220 may search an intent block by expanding keywords (synonyms, abbreviations, full names, etc.) that are semantically similar to the search term through a deep matching search model. The processor 220 uses search terms used when searching documents for documents belonging to the intent block, keywords that frequently appear in the document, categories or tags set by the user for the document, etc., to create extended keywords related to the block. can be collected. Referring to FIG. 12, for each intent block 1201 created through a naming process, an extended keyword 1210 can be collected using a document 1202 pre-matched to the corresponding block 1201. The search server 330 can search for an intent block corresponding to a search term by searching the name of the intent block as well as the extended keyword 1210 related to each block. For block search using extended keywords 1210, a neural network model (e.g., SPLADE, etc.) that learns BERT-based sparse expressions for keywords and documents can be used as a deep matching search model.

Referring again to FIG. 11, in step S1120, the processor 220 may determine the ranking of the intent block corresponding to the search word. The processor 220 may determine the ranking of the intent block based on the user's profile information and history information to provide personalized search results. For personalization, information that can identify the user's tastes (interests or preferences, etc.) can be used. Profile information can include the user's gender, age, region, etc., and history information can include search history, shopping history, and content consumption. May include history, etc. The processor 220 includes a user feature that is an embedding vector for user information (profile information and history information), a search term feature that is an embedding vector for a search word, and a block feature (topic feature and genre feature) that is an embedding vector for an intent block. You can perform personalized ranking of intent blocks corresponding to search words using a ranking model using (included).

In step S1130, the processor 220 may display personalized search results to the user by sorting intent blocks corresponding to the search term according to the ranking determined in step S1120. The processor 220 may provide search results by sorting intent blocks corresponding to search terms with a ranking that reflects user preferences.

The processor 220 may classify and provide search results corresponding to the search term into various types of blocks. The processor 220 may provide a correct answer block containing correct answer search results for a search term. The correct answer block is one of the search results corresponding to the search term and may include at least one document that is automatically determined according to logic. The processor 220 may provide a search block that includes customized search results based on user tastes. The search block is a search result that reflects the user's search intent as well as the user's taste, and may include an intent block selected through deep matching and personalization in steps S1110 to S1120.

The processor 220 may provide responsive blocks based on user responses in addition to answer-type blocks and search-type blocks. Responsive blocks, like navigation blocks, can provide search results that reflect the user's search intent and tastes, and retrieve and display optimal search results in real time by referring to the documents the user clicked on in the search results and previous search history. You can. While the answer-type block and search-type block are pre-generated blocks in which the documents to be exposed through the block are matched in advance, the responsive block is a block in which the document to be exposed through the block is determined in real time based on the clicked document. This is a real-time generation block. For example, it is possible to create and provide related responsive blocks in real time by identifying the user's interests and preferences extracted from documents that the user has clicked or reacted to for a long time. At this time, the processor 220 can dynamically collect and provide documents related to the block, or select and provide a block suitable for the user's response from blocks that have already been created.

The correct answer block is a search result that is automatically determined based on logic and provided commonly to all users, while the search block and responsive block can correspond to personalized blocks that provide different search results for each user. The correct answer block can be composed of one intent block, and the search block and responsive block can be composed of two or more intent blocks to provide search results on various topics or genres. In addition to answer-type blocks, search-type blocks, and responsive blocks, expanded search results can be provided by using various criteria to classify blocks, which are search result units. For example, documents for other search terms that are directly or indirectly related to the user's search term (e.g., search terms used sequentially before or after the search term in the search history of other users, etc.), or searches by other users with a similar profile to the user. Search results can be expanded by using additional discovery blocks that contain documents of new, frequently searched terms.

The processor 220 may determine the sorting order of the correct answer block, search block, and responsive block according to the type of search term. In the case of search words with a strong correct answer intent, the correct answer block can be arranged at the top, and in the case of search words with a strong search intent, search blocks can be arranged at the top to provide results on various topics or genres. The processor 220 can provide expanded search results through a search-type block that reflects user preferences and a responsive block that reacts in real time to the user's document clicks. Additionally, the processor 220 may exclude at least one of a correct answer block, a search block, and a responsive block from being exposed according to a search word or user preference. For example, in the case of a search term with a strong search intent, the correct answer block can be omitted and search results can be provided with only the search block and responsive block.

Accordingly, the processor 220 can provide search results optimized for the user's search intent by classifying and combining the intent blocks corresponding to the search term according to characteristics such as answer-type blocks, search-type blocks, and response-type blocks.

Figures 13 to 19 show examples of user interface screens that display search results in intent block units according to an embodiment of the present invention.

Referring to FIG. 13 , the processor 220 may classify and provide search results 1300 corresponding to a search term into correct answer blocks 1310, search blocks 1320, and responsive blocks 1330.

The search result 1300 may be composed of intent blocks selected through block search using search words.

The correct answer block 1310 may be composed of an intent block to which a correct answer document is matched. For example, the correct answer block 1310 may be composed of one intent block, and the top document may be exposed based on document ranking.

The search block 1320 may be composed of an intent block selected through personalized ranking using the user's profile information and history information. The search block 1320 may be composed of an intent block selected through personalized ranking among intent blocks corresponding to the search term entered by the user. For example, the processor 220 may classify and display the top 10 intent blocks as search blocks 1320.

The responsive block 1330 may be composed of an intent block that matches a document in real time according to a document click in the navigation block 1320. The processor 220 may recommend new results that may be of interest to the user through a graph neural network model (eg, GCN model, etc.). To this end, the relationship between search words and documents can be built into a graph, then nodes on the graph can be embedded through a graph neural network, and documents similar to the document clicked by the user can be recommended through the responsive block 1330. The responsive block 1330 may be composed of one intent block, and at least one document similar to the document clicked by the user may be exposed in the navigation block 1320.

Alternatively, the responsive block 1330 may additionally identify the user's intent based on the user's click on the document and may be composed of blocks related to this among navigational blocks that have already been created. For example, when a user searches for the query 'Jeju Island Travel' and the document clicked is related to 'Jeju Island Emotional Accommodation', if a navigational block related to 'Jeju Island Emotional Accommodation' has been created in advance, that block is selected as a responsive block. It can be provided as (1330).

Referring to FIG. 14, the processor 220 may display search results by selectively combining at least one of the answer block 1310, search block 1320, and response block 1330 according to the search word. . For example, in the case of search word i, the search results corresponding to search word i are classified and displayed into correct answer block 1310, search block 1320, and responsive block 1330, while in the case of search word ii, the correct answer block 1310 is displayed. The block 1310 can be omitted and the search results corresponding to the search term ii can be classified and displayed into a search block 1320 and a responsive block 1330.

The processor 220 is also capable of adjusting the sorting order of the answer block 1310, search block 1320, and response block 1330 according to the search word.

In particular, the processor 220 may display the searchable block 1320 as a personalized search result. The processor 220 can select a plurality of intent blocks from among the intent blocks corresponding to the search word by ranking using the user's profile information and history information, and at this time, classifies the selected intent blocks into search blocks 1320. It can be displayed as follows. For example, as shown in FIG. 15, the search block 1320 according to user I's personalization for the same search term may be composed of intent blocks Block_A, N, The corresponding search block 1320 may be composed of intent blocks Block_E, Q, B, and A.

The processor 220 may expose documents pre-matched to the corresponding block through each intent block constituting the search block 1320. Depending on the type or source of the document matched to the intent block, the number of documents exposed through the block, display method (text list type, image type, etc.), and interface method (up and down scroll, left and right swipe, etc.) can be determined. there is. As shown in FIG. 16, the number, display method, interface method, etc. of documents 1621 exposed to the corresponding block may be configured differently for each intent block 1620 constituting the search block 1320. Additionally, the method of exposing the intent block can also be configured differently for each individual, taking into account the user's response or preference. For example, for users who seem to respond more to or prefer an image-type configuration that swipes left and right rather than a text list-type that scrolls up and down, you can configure and expose an intent block by giving higher weight to that method. .

In addition to personalizing the search block 1320, each intent block constituting the search block 1320 can be personalized. It is also possible to personalize each intent block by applying ranking using the user's profile information and history information to documents pre-matched to the intent block.

The processor 220 may display a responsive block 1330 that includes a real-time search result based on user response along with a search block 1320 as a personalized search result. For example, referring to FIG. 17, when a user clicks on a specific document 1701 among the documents exposed through the navigation block 1320, the processor 220 selects a document 1700 similar to the clicked document 1701 as a responsive It can be exposed through block 1330. Intents that are not included in the search term entered by the user can be predicted through the document 1701 that the user clicks on, and the document 1700 of the intent predicted from the click document 1701 can be predicted through the responsive block 1330. Recommendations can be made in real time. In other words, the document 1700 on the responsive block 1330 can be determined through a graph neural network-based recommendation model that uses search words and clicked documents as ranking features.

The processor 220 may display search results by selectively combining at least one of the answer block 1310, search block 1320, and response block 1330 according to the search word.

For example, referring to FIG. 18, when a correct answer query such as 'YouTube' is input, the processor 220 displays a correct answer block 1810 containing a correct answer document for the query 'YouTube' at the top. You can. The processor 220 may display a search block 1820 that includes a customized search result based on user taste for the query 'YouTube' following the correct answer block 1810. At this time, the processor 220 may search for the intent block corresponding to the query 'YouTube' by searching the name of the intent block and the extended keyword related to each block among the pre-created intent blocks. Additionally, the processor 220 may personalize the search block 1820 by selecting some intent blocks among the intent blocks corresponding to the query 'YouTube' through ranking using the user's profile information and history information. For each intent block constituting the search block 1820, part of the document matched to the corresponding block is exposed, and when a user requests to view more, part of the document can be additionally exposed or the entire document can be exposed. Additionally, the processor 220 can recommend a search topic related to the query 'YouTube', and at this time, can configure a topic recommendation list 1801 using the name of the intent block corresponding to the query 'YouTube'. Depending on the embodiment, the topic recommendation list 1801 may be constructed using only intent blocks selected through personalized ranking. When a user selects a specific topic from the topic recommendation list 1801, the view area of the screen can be moved to the intent block of the selected topic and the document matching the block can be displayed. When the user selects 'Popular Channel' among the recommended topics, the search results screen may automatically scroll to the location of the intent block 'Popular Channel'.

Referring to FIG. 19, when a search-type query such as 'Jeju Island travel' is input, the processor 220 omits the correct answer block and creates a search-type block that includes customized search results based on user taste for the query 'Jeju Island travel'. (1920) can be displayed. At this time, the processor 220 may search for an intent block corresponding to the query 'Jeju Island travel' by searching the name of the intent block and extended keywords related to each block among the pre-created intent blocks. Additionally, the processor 220 may personalize the search block 1920 by selecting some intent blocks among the intent blocks corresponding to the query 'Jeju Island travel' through ranking using the user's profile information and history information. For each intent block constituting the search block 1920, part of the document matched to the corresponding block is exposed, and when a user requests to view more, part of the document can be additionally exposed or the entire document can be exposed. In addition, the processor 220 can recommend a search topic related to the query 'Jeju Island travel', and at this time, the topic recommendation list 1901 can be configured using the name of the intent block corresponding to the query 'Jeju Island travel'. . Depending on the embodiment, the topic recommendation list 1901 may be constructed using only intent blocks selected through personalized ranking. When a user selects a specific topic from the topic recommendation list 1901, the intent block of the selected topic can be accessed and a document matching the block can be provided. If the user selects 'Jeju Island travel destination' among the recommended topics, the search results screen may automatically scroll to the location of the intent block 'Jeju Island travel destination'.

The processor 220 includes a search block 1920 that includes customized search results based on user preferences for the query 'Jeju Island travel' as a personalized search result, and a responsive block (Mido) that includes real-time search results based on user reaction. hour) can be displayed. When a user clicks on a specific document among the documents exposed through the search block 1920 for the query 'Jeju Island travel', the processor 220 may expose a document similar to the clicked document through a responsive block. Accordingly, the processor 220 may provide expanded search results by additionally exposing documents searched in real time based on the document clicked on in the search block 1920.

Meanwhile, by creating an intent block that matches the user's intent and providing search results in block units, content creators can be informed of what kind of content to create, which has the effect of encouraging content creation. Referring to FIG. 20, in one embodiment of the present invention, a participation button 2001 that can create content (document) related to the intent block 2000 is added at a location associated with each intent block 2000. can be provided. At this time, in order to improve the quality of the search target content, the participation button 2001 for the intent block 2000 can be exposed only to users who have already been verified or who have produced a lot of content related to the corresponding intent block. Content created in this way can be directly reflected in the document pool of the corresponding intent block (2000). If necessary, it can be ranked and exposed along with general content, or only content created through participation can be separately exposed.

As such, according to embodiments of the present invention, it is possible to provide search results that take individual intent and taste into account using blocks, which are new search result units that subdivide search intents. According to embodiments of the present invention, topics and genres can be extracted through document classification, and blocks in units of search intents can be created by combining topics and genres rather than the type or source of content. According to embodiments of the present invention, a block can be created for each keyword with a combination of topics and genres corresponding to the keyword, and then documents to be exposed through the block can be matched to construct data that can be used for real-time search. According to embodiments of the present invention, search results can be provided in personalized blocks by considering the user's search intent based on the search word and the user's personal preferences according to the context. According to embodiments of the present invention, a correct answer block containing correct answer search results according to search intent, a search block containing customized search results based on user taste, and a response containing real-time search results based on user reaction. Type blocks can be freely combined to provide optimized search results.

The device described above may be implemented with hardware components, software components, and/or a combination of hardware components and software components. For example, the devices and components described in the embodiments include a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), and a programmable logic unit (PLU). It may be implemented using one or more general-purpose or special-purpose computers, such as a logic unit, microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device includes multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include a plurality of processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are possible.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. The software and/or data may be embodied in any type of machine, component, physical device, computer storage medium or device for the purpose of being interpreted by or providing instructions or data to the processing device. there is. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on 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 on a computer-readable medium. At this time, the medium may continuously store a computer-executable program, or temporarily store it for execution or download. In addition, the medium may be a variety of recording or storage means in the form of a single or several pieces of hardware combined. It is not limited to a medium directly connected to a computer system and may be distributed over a network. Examples of media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, And there may be something configured to store program instructions, including ROM, RAM, flash memory, etc. Additionally, examples of other media include recording or storage media managed by app stores that distribute applications, sites or servers that supply or distribute various other software, etc.

As described above, although the embodiments have been described with limited examples and drawings, various modifications and variations can be made by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

Claims (20)

In a personalized search method executed on a computer device,
The computer device includes at least one processor configured to execute computer-readable instructions contained in a memory,
The personalized search method is,
generating a search intent unit block using a combination of a topic and a genre by the at least one processor; and
Providing personalized search results using the block selected based on context related to the user for a search term given by the user, by the at least one processor.
Personalized search method including. According to paragraph 1,
The generating step is,
extracting the topic and the genre through document analysis of documents on the Internet;
determining a naming of the block, which is a search result unit, by combining the topic and the genre; and
Connecting a document to be exposed as a search result through the block for each of the blocks.
Personalized search method including. According to paragraph 1,
The generating step is,
Selecting keywords for block creation;
generating the block, which is a search result unit, by combining the topic and the genre for each of the keywords; and
Connecting a document to be exposed as a search result through the corresponding block for each of the blocks.
Personalized search method including. According to paragraph 3,
The generating step is,
Periodically updating the block information for each keyword and the document information for each block in a cache for real-time search.
Personalized search method including. According to paragraph 1,
The generating step is,
Extracting the topic for intent classification through topic modeling for documents on the Internet;
extracting the genre to subdivide the topic by classifying the document according to document format;
determining a naming of the block, which is a search result unit, by combining the topic and the genre; and
Connecting a document to be exposed as a search result through the block for each of the blocks.
Personalized search method including. According to paragraph 1,
The steps provided above are:
Searching for a block corresponding to the search word among the blocks; and
Determining the ranking of the block corresponding to the search term based on at least one user information selected from the user's profile information and history information.
Personalized search method including. According to clause 6,
The above search step is,
Searching for blocks whose names contain keywords that match the search term or keywords that are semantically similar to the search term.
A personalized search method featuring: According to clause 6,
The above search step is,
Searching for blocks of names containing keywords corresponding to the search term using a deep matching model based on a neural network that learns sparse expressions for keywords and documents.
A personalized search method featuring: According to clause 6,
The above search step is,
collecting extended keywords related to the block using documents pre-matched to the block; and
Searching for a block corresponding to the search term using the name of the block and the extended keyword
Personalized search method including. According to clause 6,
The determining step is,
Personalized ranking of blocks corresponding to the search term using a ranking model using a user feature representing an embedding vector for the user information, a search term feature representing an embedding vector for the search term, and a block feature representing an embedding vector for the block. to do
A personalized search method featuring: A computer program stored in a computer-readable recording medium for executing the personalized search method of any one of claims 1 to 10 on a computer. In computer devices,
At least one processor configured to execute computer readable instructions contained in memory
Including,
The at least one processor,
Create a search intent-level block using a combination of topic and genre,
Providing personalized search results for a search term given by a user using the block selected based on context related to the user.
A computer device characterized by a. According to clause 12,
The at least one processor,
Extract the topic and genre through document analysis of documents on the Internet,
Determine the naming of the block, which is a search result unit, by combining the topic and the genre,
Connecting documents to be exposed as search results through the corresponding block for each of the blocks above.
A computer device characterized by a. According to clause 12,
The at least one processor,
Select keywords for block creation,
Generating the block, which is a search result unit, by combining the topic and the genre for each of the keywords,
Connecting documents to be exposed as search results through the corresponding block for each of the blocks above.
A computer device characterized by a. According to clause 14,
The at least one processor,
Periodically updating the block information for each keyword and the document information for each block in a cache for real-time search
A computer device characterized by a. According to clause 12,
The at least one processor,
Search for a block corresponding to the search word among the blocks,
Determining the ranking of the block corresponding to the search term based on at least one user information of the user's profile information and history information.
A computer device characterized by a. According to clause 16,
The at least one processor,
Searching for blocks whose names contain keywords that match the search term or keywords that are semantically similar to the search term.
A computer device characterized by a. According to clause 16,
The at least one processor,
Searching for blocks of names containing keywords corresponding to the search term using a deep matching model based on a neural network that learns sparse expressions for keywords and documents.
A computer device characterized by a. According to clause 16,
The at least one processor,
Collect extended keywords related to the block using documents pre-matched to the block,
Searching for a block corresponding to the search term using the name of the block and the extended keyword
A computer device characterized by a. According to clause 16,
The at least one processor,
Personalized ranking of blocks corresponding to the search term using a ranking model using a user feature representing an embedding vector for the user information, a search term feature representing an embedding vector for the search term, and a block feature representing an embedding vector for the block. to do
A computer device characterized by a.

Images