KR102082202B1 - Device and method for providing recommended information on artificial intelligence based customized tools - Google Patents

Abstract

The present invention relates to an artificial intelligence (AI) based apparatus for providing customized tool recommendation information and a method thereof. According to an embodiment of the present invention, an AI based apparatus for providing customized tool recommendation information comprises: a communication unit for receiving tool product purchase information and tool product history information for tool product purchase from a user terminal; a pattern calculating unit for calculating a purchase pattern and a history pattern from the tool product purchase information and the tool product history information; and a recommendation unit which generates customized tool recommendation information for a new tool product purchase request based on the purchase pattern and the history pattern. The communication unit may transmit the customized tool recommendation information to the user terminal.

Description

DEVICE AND METHOD FOR PROVIDING RECOMMENDED INFORMATION ON ARTIFICIAL INTELLIGENCE BASED CUSTOMIZED TOOLS}

The present invention relates to an apparatus and method for providing AI-based customized tool recommendation information.

The tool is used in various construction or industrial sites, and is one of tools that are closely utilized in daily life.

In addition to commonly used bolts and nuts, there are various types of tools, and even the same type may be provided in different sizes depending on the purpose. Accordingly, even if you purchase the same tool as the previously purchased tool, there is a fragment point that the specifications of the type, value, etc. of the tool should be provided to the seller at every purchase.

The background technology of the present application is disclosed in Korean Patent Publication No. 10-2016-0095477.

The present invention is to solve the above-mentioned problems of the prior art, an object of the present invention to provide an artificial intelligence-based customized tool recommendation information providing apparatus and method that provides a customized tool recommendation information by learning the purchase history of the user.

The present application is to solve the above-mentioned problems of the prior art, an artificial intelligence-based customized tool recommendation information providing apparatus for recognizing text input to the user terminal, learning the contents of the tool product purchase to provide customized tool recommendation information and The purpose is to provide a method.

However, the technical problem to be achieved by the embodiments of the present application is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, the artificial intelligence-based customized tool recommendation information providing apparatus according to an embodiment of the present application, receiving the tool product purchase information and tool product details information for the tool product purchase from the user terminal Communication unit; A pattern calculator configured to calculate a purchase pattern and a detail pattern from the tool product purchase information and the tool product detail information; And a recommendation unit generating customized tool recommendation information for a new tool product purchase request based on the purchase pattern and the detail pattern, wherein the communication unit may transmit the customized tool recommendation information to the user terminal.

According to the exemplary embodiment of the present application, the tool product purchase information may include purchase time information of the tool, and the tool product detail information may include at least one of type information of the tool and dimension information for each tool.

According to the exemplary embodiment of the present application, the pattern calculating unit may construct an unsupervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of the individual user as the input and the purchase pattern of the individual user. The history pattern may be calculated, and the purchase pattern may include a pattern for each month, week, and time, and the history pattern may include a pattern for each tool type and a pattern for each dimension.

According to one embodiment of the present application, the pattern calculation unit, the non-supervised learning-based pattern calculation model for inputting the tool product purchase information and the tool product details information of all the user by the purchase pattern of the entire user and The detail pattern is calculated, and the recommendation unit may generate customized tool recommendation information for the new tool product request based on the purchase pattern and the detail pattern of the entire user.

According to one embodiment of the present application, the pattern calculation unit, the natural language processing by collecting the text input to the user terminal through the communication unit, the purchase request according to the tool product purchase based on the text and the surrounding words of the text The information may be generated, and the purchase pattern and the detail pattern may be calculated based on the purchase request information.

According to an embodiment of the present disclosure, if the number of preset reference texts in the natural language-processed text is greater than or equal to a preset threshold, it is determined that the text is associated with the purchase of the tool product, and the surrounding words are associated with the reference text. It may be selected when the degree is greater than or equal to a preset threshold.

According to an embodiment of the present application, the pattern calculator may form a noun extracted from the text through a morphological analysis into a vector space, and select the surrounding word based on a distance between vectors of the noun.

According to one embodiment of the present application, the type of tool, hex bolt, heavy hex screw, hex cap screw, U bolt, set anchor, all anchor, wing bolt, eye bolt, hinge bolt, hex nut, square nut, wing At least one of a nut, an eye nut, and a tab washer may be included.

According to an embodiment of the present application, the tool dimension information includes a specification for each type of the tool, and the recommendation unit constructs a dimension recommendation model based on unsupervised learning based on the tool dimension information according to the specification for each tool type. The customized tool recommendation information according to the dimension-specific pattern may be generated.

Artificial intelligence-based customized tool recommendation information providing method according to an embodiment of the present invention, receiving the tool product purchase information and tool product details for the purchase of the tool product from the user terminal; Calculating a purchase pattern and a detail pattern from the tool product purchase information and the tool product detail information; Generating customized tool recommendation information for a new tool product purchase request based on the purchase pattern and the history pattern; And transmitting the customized tool recommendation information to the user terminal.

According to one embodiment of the present application, the tool product purchase information includes purchase time information of the tool,

The tool product detail information may include at least one of the type information of the tool and the dimension information for each tool.

According to an embodiment of the present disclosure, the calculating of the purchase pattern and the detail pattern includes constructing a non-supervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of an individual user as the input. The purchase pattern and the detail pattern of the individual user may be calculated, wherein the purchase pattern includes monthly, weekly, and hourly patterns, and the detail pattern may include a pattern by tool type and a pattern by dimension.

According to an exemplary embodiment of the present disclosure, the calculating of the purchase pattern and the detail pattern may include constructing a non-supervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of all users as input. The calculating of the purchase pattern and the detail pattern of all users and generating the customized tool recommendation information may include generating customized tool recommendation information for the new tool product request based on the purchase pattern and the history pattern of all users. can do.

According to an exemplary embodiment of the present disclosure, the calculating of the purchase pattern and the detail pattern may include collecting text input to the user terminal and processing natural language, and based on the purchase of a tool product based on the text and surrounding words of the text. The purchase request information may be generated, and the purchase pattern and the detail pattern may be calculated based on the purchase request information.

According to an embodiment of the present disclosure, the calculating of the purchase pattern and the history pattern may include determining that the text is associated with the tool product purchase when the number of reference texts preset in the natural language-processed text is greater than or equal to a preset threshold. The peripheral word may be selected when the degree of association with the reference text is greater than or equal to a preset threshold.

According to an embodiment of the present disclosure, the calculating of the purchase pattern and the detail pattern may include forming a noun extracted from the text into a vector space through morphological analysis, and calculating the surrounding words based on a distance between the vectors of the noun. Can be selected.

According to one embodiment of the present application, the type of tool, hex bolt, heavy hex screw, hex cap screw, U bolt, set anchor, all anchor, wing bolt, eye bolt, hinge bolt, hex nut, square nut, wing At least one of a nut, an eye nut, and a tab washer may be included.

According to an embodiment of the present application, the tool numerical information includes a specification for each type of the tool, and the step of generating the customized tool recommendation information, non-supervised learning to input the tool dimension information according to the specification for each tool type The customized tool recommendation information according to the pattern for each dimension may be generated by constructing a base dimension recommendation model.

The above-mentioned means for solving the problems are merely exemplary, and should not be construed as limiting the present application. In addition to the above-described exemplary embodiments, additional embodiments may exist in the drawings and detailed description of the invention.

According to the above-described problem solving means of the present application, it is possible to provide an artificial intelligence-based customized tool recommendation information providing apparatus and method that provides a customized tool recommendation information by learning the purchase history of the user.

According to the above-described problem solving means of the present application, to provide an apparatus and method for providing AI-based customized tool recommendation information for recognizing text input to the user terminal, learning the contents of the tool product purchase to provide customized tool recommendation information Can be.

1 is a diagram illustrating a configuration of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.
FIG. 2 is a diagram illustrating an example of a purchase pattern and a detail pattern detection of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.
3 is a view showing the specifications of the hex bolt of the artificial intelligence-based customized tool recommendation information providing apparatus according to an embodiment of the present application.
4 is a view showing the specifications of the heavy hex screw of the artificial intelligence-based customized tool recommendation information providing apparatus according to an embodiment of the present application.
FIG. 5 is a view illustrating specifications of an all anchor of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.
6 is a diagram illustrating a hierarchical structure of WordNet for text mining of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.
7 is a diagram illustrating an example of customized tool recommendation information output from a user terminal by an apparatus for providing AI-based customized tool recommendation information according to an embodiment of the present disclosure.
8 is a flowchart illustrating a method of providing AI-based customized tool recommendation information according to an exemplary embodiment of the present application.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present disclosure. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted for simplicity of explanation, and like reference numerals designate like parts throughout the specification.

Throughout this specification, when a part is "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element in between. do.

Throughout this specification, when a member is said to be located on another member "on", "upper", "top", "bottom", "bottom", "bottom", this means that any member This includes not only the contact but also the case where another member exists between the two members.

Throughout this specification, when a part is said to "include" a certain component, it means that it can further include other components, without excluding the other components unless otherwise stated.

1 is a diagram illustrating a configuration of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.

Referring to FIG. 1, the apparatus 100 for providing artificial intelligence-based customized tool recommendation information may include a communicator 110, a pattern calculator 120, and a recommender 130. The communication unit 110 may receive tool product purchase information and tool product detail information for purchasing a tool product from the user terminal 200. The tool product purchase information may be generated when a user purchases a tool through a tool sales application or a website using the user terminal 200. Specifically, the tool product purchase information may include purchase time information of the tool. The tool product detail information may include at least one of the type information of the purchased tool and the dimension information for each tool.

According to one embodiment of the present application, the tool product purchase information and the tool product details information, as well as the tool sales application or website, the user through the E-mail, text message, MMS, messenger application using the user terminal 200 It can also be created if you buy the tool directly from the seller. Tool product purchase information and tool product detail information generated by purchasing directly to the seller will be described later.

The communication unit 110 may receive a payment history of a tool product purchase from the user terminal 200, and the pattern calculation unit 120 may collect time information, tool type information, and tool-specific dimension information from the payment history. Can be. In addition, the communication unit 110 may collect text input to the user terminal 200, and the pattern calculator 120 may calculate time information, tool type information, and tool numerical information from the text. Information calculated from the text will be described later.

The apparatus 100 for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present application may communicate with the user terminal 200 through a network. The network refers to a connection structure capable of exchanging information between respective nodes such as terminals and servers. Examples of such a network include a 3rd generation partnership project (3GPP) network, a long term evolution (LTE) network, and 5G. Network, World Interoperability for Microwave Access (WIMAX) network, Internet (Internet), Local Area Network (LAN), Wireless Local Area Network (WLAN), Wide Area Network (WAN), Personal Area Network (PAN), wifi network, Bluetooth networks, satellite broadcasting networks, analog broadcasting networks, digital multimedia broadcasting (DMB) networks, etc. may be included, but is not limited thereto. In addition, the user terminal 200 may include, for example, a smartphone, a smart pad, a tablet PC, a personal communication system (PCS), a global system for mobile communication (GSM), a personal digital cellular (PDC), Personal Handyphone System (PHS), Personal Digital Assistant (PDA), International Mobile Telecommunication (IMT) -2000, Code Division Multiple Access (CDMA) -2000, W-Code Division Multiple Access (W-CDMA), Wireless Broadband Internet ) May include all kinds of wireless communication devices such as terminals, and wired communication devices such as desktop computers and smart TVs.

The pattern calculator 120 may calculate a purchase pattern and a detail pattern from the tool product purchase information and the tool product detail information. The pattern calculator 120 calculates the purchase pattern and the detail pattern of the individual user by constructing a non-supervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of the individual user as input. Can be. The pattern calculator 120 may form a tool product purchase information and the tool product detail information and use it as an input of a pattern calculation model based on unsupervised learning. For example, in the case of tool product purchase information, the standardization means to digitize a tool order date, a tool order time, a tool order period, and the like. In addition, in the case of tool product detail information, it means that the type of tool and the dimension for each tool are digitized. For example, the pattern calculator 120 may assign a numerical value according to each type of tool.

The pattern calculator 120 may calculate a purchase pattern and a breakdown pattern based on an unsupervised learning based clustering algorithm. Unsupervised learning refers to an algorithm that learns by analyzing or clustering data itself, rather than constructing learning data. The pattern calculator 120 may cluster and calculate a purchase pattern and a detail pattern based on a clustering algorithm, and detect a new purchase pattern and a detail pattern based on a separation degree between the purchase pattern and the detail pattern. For example, a logistic regression algorithm, a random forest algorithm, a support vector machine (SVM) algorithm, a decision algorithm, and a clustering algorithm may be used as the clustering algorithm for unsupervised learning. In addition to the above-described algorithm, the new pattern classification unit 130 may use clustering algorithms such as Extra Tree algorithm, XG Boost algorithm, Deep Learning algorithm, K-means clustering algorithm, Self-Organizing-Maps (SOM) algorithm, EM & Canopy algorithm. Can lead to unsupervised learning. The random forest algorithm is an algorithm that makes a number of decision trees to form a forest and averages each prediction result as a single result variable. The SVM algorithm determines the classification to which data belongs by dividing the largest boundary in the data distribution space. It is a non-probability algorithm. The Extra Tree algorithm is similar to the Random forest, but the algorithm is faster than the Random forest. The XGBoost algorithm is a boost algorithm that applies the result of the XGBoost tree to the next tree if the tree of the Random Forest is independent. Deep Learning algorithm is an algorithm that learns by adjusting the effect of the pattern of variable on the weight based on multi-layer Neural Network. In addition, K-means clustering algorithm is a traditional classification technique that repeats the target group into K clusters based on the mean value (similarity) of distance, and the SOM algorithm is based on the artificial neural network. It is a technique of learning and clustering. In addition, EM & Canopy algorithm refers to a technique for clustering by updating the parameter value through the iterative process from the maximum probability to the given initial value.

FIG. 2 is a diagram illustrating an example of a purchase pattern and a detail pattern detection of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.

The pattern calculator 120 detects a purchase pattern through an unsupervised learning-based pattern calculation model using the tool product purchase information, and generates a non-supervised learning-based pattern calculation model using the tool product details information as an input. The history pattern can be detected. In other words, the purchase pattern and the detail pattern may be detected based on unsupervised learning using the tool product purchase information and the tool product detail information according to the past history of the individual user. In addition, unsupervised learning for detecting a purchase pattern and unsupervised learning for detecting a history pattern may be performed respectively. FIG. 2 illustrates a grouping history pattern according to the frequency of the tool product purchase information and the tool product history information. When the frequency of different tool purchase information and tool request information is similar, the same to similar labeling (as shown in FIG. Blue and green). That is, when the labeling between the history pattern or the purchase pattern is the same or similar, the degree of separation between clusters may be low. In other words, different history patterns or purchase patterns may be detected through respective clusters according to the frequency of the tool purchase information and the tool request information.

Meanwhile, in the case of a purchase pattern (or a detail pattern) clustered with tool product purchase information and tool product history information having a different frequency from tool product purchase information and tool product history information having a similar frequency, as shown in FIG. 2. As described above, it may be distinguished from a labeling (red) different from the above-described purchase pattern. Since the purchase pattern is different from the frequency of the tool product purchase information and the tool product detail information of another purchase pattern, it may be a new purchase pattern (or a detail pattern) and may be detected by the pattern calculator 120. The purchase pattern detected by the pattern calculator 120 may include a pattern for each month, week, and time, and the detail pattern detected by the pattern calculator 120 may include a pattern for each tool type and a pattern for each dimension. Exemplary types of tools are hex bolt, heavy hex screw, hex cap screw, U bolt, set anchor, all anchor, wing bolt, eye bolt, hinge bolt, hex nut, square nut, wing nut, eye nut, tab It may include at least one of the washers. According to one embodiment of the present application, not only the type of the tool described above, but also a pattern by a tool type and a pattern by dimension for various kinds of tools may be calculated.

The recommendation unit 130 may generate customized tool recommendation information for a new tool product purchase request based on the purchase pattern and the detail pattern. For example, when a purchase pattern and a history pattern of an individual user are calculated, the next tool purchase date may be predicted by the purchase pattern, and the type and purpose of the customized tool to be purchased next may be determined by the history pattern to determine tool recommendation information. Can be generated as In other words, the recommendation unit 130 may generate customized tool recommendation information before requesting a new tool product purchase based on the purchase pattern and the history pattern and transmit the customized tool recommendation information to the user terminal 200 through the communication unit 110. In addition, the recommendation unit 130 may generate customized tool recommendation information including tool dimension information of the customized tool which is recommended based on the purchase pattern and the detail pattern. The tool dimension information may include specifications for each type of tool.

3 is a view showing the specifications of the hex bolt of the artificial intelligence-based customized tool recommendation information providing apparatus according to an embodiment of the present application, Figure 4 is an artificial intelligence-based customized tool recommendation information providing apparatus according to an embodiment of the present application FIG. 5 is a diagram illustrating specifications of a heavy hex screw, and FIG. 5 is a diagram illustrating specifications of an all anchor of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment.

Each tool has different specifications according to the shape of the tool for each type. For example, referring to FIG. 3, hex bolts may include body dia, width across flats, width across corners, head height, fillet radius ( radius of fillet, tread length specifications. In addition, the size of each specification may be different for each arc (nominal size) of the hex bolt. On the other hand, referring to Figure 5, the all anchor (all anchor) may include the specifications of the total length (thread length), the thread length (thread length), tensile strength (tensile strength), by the number of arc by all anchors The size may be different. As described above, since the types and values of the specifications are different for each tool, the recommendation unit 130 constructs a non-supervised learning-based dimension recommendation model that inputs the tool dimension information according to the specifications for each tool type, and thus the Create custom tool recommendations. For example, when a purchase pattern and a detail pattern are generated by periodically purchasing tools of the same size, the recommendation unit 130 may generate customized tool recommendation information about tools of the same type and dimension as before. Based on the same cycle, before the next tool purchase date arrives, it may be transmitted to the user terminal 200 to provide customized tool recommendation information.

The recommendation unit 130 may recognize the connection of the tool sales application or website as a tool purchase request. The customized tool recommendation information generated by the recommendation unit 130 may be stored in the user terminal 200 through the communication unit 110. When the user accesses a tool sales application or a website, the user terminal 200 is new. Recognizing the intention to purchase a tool product, the stored customized tool recommendation information may be output. As another example, when the user accesses the tool sales application or website, the user terminal 200 transmits whether the tool sales application or website is connected to the communication unit 110, and the recommendation unit 130 is a tool sales application or Customized tool recommendation information for requesting a new tool product purchase request according to the access of a website may be generated in real time and transmitted to the user terminal 200 through the communication unit 110.

As described above, the purchase pattern and the detail pattern detected based on the tool product purchase information and the tool product detail information of the user use the past history of the user as input data for learning. However, a new user whose past history is not accumulated enough to detect a pattern or does not have a past history cannot generate a purchase pattern and a history pattern through the individual's past history. Accordingly, the pattern calculator 120 may calculate a purchase pattern and a detail pattern of all users by constructing a non-supervised learning-based pattern calculation model using inputs of tool product purchase information and tool product details of all users. In addition, the recommendation unit 130 may generate customized tool recommendation information for a new tool product request based on the purchase pattern and the history pattern of all users.

6 is a diagram illustrating a hierarchical structure of WordNet for text mining of an apparatus for providing AI-based customized tool recommendation information according to an exemplary embodiment of the present disclosure.

As described above, the communication unit 110 may collect text input to the first user terminal 200, and the pattern calculator 120 calculates time information, tool type information, and tool numerical information from the text. can do. The pattern calculator 120 collects the text input to the user terminal through the communication unit 110 and processes the natural language, and the tool product purchase information and the tool according to the tool product purchase and tool processing request based on the text and surrounding words of the text. Processing request information can be generated. In addition, the pattern calculator 120 may calculate a purchase pattern and a breakdown pattern based on the tool product purchase information and the tool processing request information.

Specifically, natural language processing refers to an understanding of natural language that mechanically analyzes language phenomena spoken by humans and makes it into a form that can be understood by a computer, or various technologies that express such a form in a language that can be understood by humans. This natural language processing can be done through morphological analysis and part-of-speech attachment. Morphological analysis, for example, 'composite nouns' can be broken up in various ways, such as 'compound + noun', 'compound + noun', 'compound noun + sa', etc. In order to select a suitable result from the various analysis results, a dynamic programming method called table parsing is used. Specifically, if the probability of splitting into a compound + noun is greater than the probability of being analyzed elsewhere, it is an algorithm that selects it. Part-of-speech is the attachment of the part of speech that is most appropriate from the results of morphological analysis. Modules, commonly called taggers, perform this function. This is a module that selects one analysis result suitable for the context from various analysis results outputted by the morpheme analyzer. In the analysis, the appropriate analysis results are selected using information that hints of neutrality located at the left and right of the context. Normally, taggers are implemented using large parts of speech. Hidden Hidden Markov Models (HMMs) are widely used.

The pattern calculator 120 may recognize text of the user terminal 200 through natural language processing, and extract text related to a tool from the text. For example, the pattern calculator 120 may determine that the text is associated with the tool product purchase when the number of preset reference texts in the natural language-processed text is greater than or equal to a preset threshold. The reference text may be set as text related to ordering or purchasing a tool product. For example, names of tools such as bolts, nuts, anchors, washers, and the like, months, days, hours, etc., keywords related to the purchase of the tool may be set as the reference text.

When the reference text is greater than or equal to a predetermined threshold value, since there are a number of keywords related to the tool among the texts processed in natural language, it may be considered that there is an intention to order the tool by using a text message, a messenger, an E-mail, and the like. However, since the reference text alone cannot accurately classify the specific order date or the type or dimension of the tool, the pattern calculator 120 extracts the surrounding words, which are terms related to the reference text, to purchase a specific tool product or request for tool processing. Tool product purchase information and tool processing request information including a may be generated. When selecting the surrounding words of the reference text, it is necessary to grasp the characteristic words, not any words, and to consider the relationship information between words. This is to prevent the occurrence of wrong tool product purchase information and tool processing request information by selecting a word that is less related to the reference text or selecting a highly related word through indiscriminate selection of surrounding words. To this end, it is necessary to select the surrounding words in consideration of the meaning between words, and it is necessary to increase the quality of the feature by adding synonyms or upper words of each of the reference text or the surrounding words. For example, WordNet, which is an example of a lexicon, may be used. The lexical dictionary is a dictionary containing the relationship information of the semantic vocabulary list of words, and is composed of synonym set units, and provides semantic relations such as upper words, lower words, equiwords, and whole words for each synonym set.

Referring to FIG. 6, according to the hierarchical structure of WordNet, Car 30 and Bicycle 40 are commonly lower words of Vehicle 10, and two words are connected to upper and lower words through node 20, respectively. have. Therefore, the distance between words can be determined according to the number of nodes connected. When the surrounding words of the reference text selected using the word relationship of WordNet are selected, the pattern calculator 120 may calculate the degree of association between the reference text and the surrounding words using a similarity function. The similarity function may be expressed as in Equation 1.

[Equation 1]

Here, a represents reference text and f represents a surrounding word. Min dist means the minimum distance in WordNet upper / lower term relationship, and common parent means common upper term of two words. Finally, root means the highest level word. This expression is a normalized expression. If two words are synonymous, the minimum distance is 0 and the correlation is 1, and if there is no common upper word, the minimum distance is infinite and the association is 0. When the degree of association between the reference text and the surrounding words is calculated by the similarity function of Equation 1, the pattern calculator 120 may select a word having a degree of association greater than or equal to a preset threshold as the surrounding words. The preset threshold may have a value between 0 and 1, which is a result of the similarity function (Equation 1).

The surrounding words that have a high degree of association with the reference text may be words that have a high degree of association with the tool. Accordingly, the pattern calculator 120 may determine whether the corresponding text is text related to a tool according to the number of reference texts, and more specifically, determine whether there is a mention of a type, a size, etc. of the tool through surrounding words. As the text is processed in natural language, numbers and numerical values such as dates and dimensions may also be recognized. As a result, the pattern calculator 120 may determine whether the text is about the purchase of the tool product through the reference text and the surrounding words, and the details such as the tool product purchase information including information about the tool product purchase and the dimensions of the tool to be purchased. The tool product detail information including the information may be generated. In addition, the tool product purchase information and the tool product detail information may be used as an input of the above-described pattern calculation model to calculate the purchase pattern and the detail pattern. The recommendation unit 130 may generate tool dimension information of the customized tool based on the purchase pattern and the history pattern calculated through the text mining, and generate the customized tool information including the tool dimension information. Through the user terminal 200 can be transmitted.

As another example, the pattern calculator 120 may select surrounding words by performing morphological analysis of the text collected from the user terminal 200. In detail, the pattern calculator 120 may format a noun extracted from the text into a vector space through morpheme analysis. For example, the pattern calculator 120 may mark text through Komoran (Korean Morphological Analyzer) morpheme analysis and extract a noun from the text. In addition, the nouns extracted by morphological analysis can be formalized by mapping them into vector space through Word2Vec, a word relationship analysis model. Unlike conventional morphological analysis, Komoran morphological analysis can analyze multiple words in a single part of speech, which enables more accurate marking of tool-related text including spaces. The pattern calculator 120 may form a word noun extracted through morphological analysis to select a surrounding word based on a distance between the vectors by converting the marked word into a vector. For example, the pattern calculator 120 may select a word having a distance between vectors less than a predetermined distance as a surrounding word.

7 is a diagram illustrating an example of customized tool recommendation information output from a user terminal by an apparatus for providing AI-based customized tool recommendation information according to an embodiment of the present disclosure.

The customized tool recommendation information generated by the recommendation unit 130 may be transmitted to the user terminal 200 through the communication unit 110, and the user terminal 200 outputs the customized tool recommendation information as illustrated in FIG. 7. can do. The customized tool recommendation information output through the user terminal 200 may include an image of a tool, a type of a tool, an arc of a type, and a specification of a type.

8 is a flowchart illustrating a method of providing AI-based customized tool recommendation information according to an exemplary embodiment of the present application.

The method for providing AI-based customized tool recommendation information illustrated in FIG. 8 may be performed by the apparatus for providing AI-based customized tool recommendation information described above with reference to FIGS. 1 to 7. Therefore, even if omitted below, the description of the apparatus for providing AI-based customized tool recommendation information through FIGS. 1 to 7 may be equally applied to FIG. 8.

Referring to FIG. 8, in operation S810, the communication unit 110 may receive tool product purchase information and tool product detail information for purchasing a tool product from a user terminal. The tool product purchase information may be generated when a user purchases a tool through a tool sales application or a website using the user terminal 200. Specifically, the tool product purchase information may include purchase time information of the tool. The tool product detail information may include at least one of the type information of the purchased tool and the dimension information for each tool. Exemplary types of tools include hex bolts, heavy hex screws, hex cap screws, U bolts, set anchors, all anchors, wing bolts, eye bolts, hinge bolts, hex nuts, square nuts, wing nuts, eye nuts. And a tap washer. The tool numerical information may include specifications for each type of the tool.

In operation S820, the pattern calculator 120 may calculate a purchase pattern and a detail pattern from the tool product purchase information and the tool product detail information. For example, the pattern calculator 120 constructs an unsupervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of the individual user as inputs, thereby providing the purchase pattern and the details of the individual user. The pattern can be calculated. In addition, the purchase pattern may include a pattern for each month, week, and time, and the detail pattern may include a pattern for each tool type and a pattern for each dimension.

In addition, the pattern calculator 120 constructs a non-supervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of all users to input the purchase pattern and the detail pattern of the entire user. Can be calculated. The recommendation unit 130 may generate customized tool recommendation information for the new tool product request based on the purchase pattern and the detail pattern of the entire user.

In addition, the pattern calculator 120 may collect the text input to the user terminal and process the natural language, and generate the purchase request information according to the purchase of the tool product based on the text and surrounding words of the text. In addition, the pattern calculator 120 may calculate the purchase pattern and the breakdown pattern based on the purchase request information.

In addition, the pattern calculator 120 may determine that the text is associated with the tool product purchase when the number of reference texts preset in the natural language-processed text is greater than or equal to a preset threshold. The surrounding word may be selected when the degree of association with the reference text is greater than or equal to a preset threshold. As another example, the pattern calculator 120 may form a noun extracted from the text into a vector space through morphological analysis, and select the surrounding word based on the distance between the vectors of the noun.

In operation S830, the recommendation unit 130 may generate customized tool recommendation information for a new tool product purchase request based on the purchase pattern and the detail pattern. The recommendation unit 130 may generate the customized tool recommendation information according to the pattern for each dimension by building a dimension recommendation model based on non-supervised learning that receives the tool dimension information according to the specifications for each tool type.

The recommendation unit 130 may recognize the connection of the tool sales application or website as a tool purchase request. The customized tool recommendation information generated by the recommendation unit 130 may be stored in the user terminal 200 through the communication unit 110. When the user accesses a tool sales application or a website, the user terminal 200 is new. Recognizing the intention to purchase a tool product, the stored customized tool recommendation information may be output. As another example, when the user accesses the tool sales application or website, the user terminal 200 transmits whether the tool sales application or website is connected to the communication unit 110, and the recommendation unit 130 is a tool sales application or Customized tool recommendation information for requesting a new tool product purchase request according to the access of a website may be generated in real time and transmitted to the user terminal 200 through the communication unit 110.

In operation S840, the communicator 110 may transmit the customized tool recommendation information to the user terminal 200. In addition, the communication unit 110 may transmit the customized tool recommendation information including the tool dimension information generated by the recommendation unit 130 to the user terminal 200.

According to an embodiment of the present disclosure, the method for providing AI-based customized tool recommendation information may be implemented in a program instruction form that can be executed by various computer means and recorded in a computer readable medium. The computer readable medium may include program instructions, data files, data structures, and the like, alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

The above description of the present application is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present application is indicated by the following claims rather than the above description, and it should be construed that all changes or modifications derived from the meaning and scope of the claims and their equivalents are included in the scope of the present application.

100: AI-based customized tool recommendation information providing device
110: communication unit
120: pattern calculation unit
130: recommendation
200: user terminal

Claims (19)

In the artificial intelligence-based customized tool recommendation information providing apparatus,
Tool product purchase information and tool type information and tool-specific information including the purchase time information of the tool product for the purchase of the tool product including nuts, bolts, anchors and washers used in the industrial field from the user terminal Communication unit for receiving the tool product details information comprising a;
A pattern for calculating a purchase pattern including monthly, weekly and hourly patterns from the tool product purchase information and the tool product detail information, and a detail pattern including a pattern for each tool type and a pattern for each dimension in consideration of different specifications according to shapes of tool types A calculator; And
A recommendation unit for generating customized tool recommendation information including tool dimension information based on specifications for each type of customized tool for a new tool product purchase request based on the purchase pattern and the history pattern,
The communication unit,
Transmit the customized tool recommendation information to the user terminal,
The pattern calculator collects text input through the E-mail, text message, MMS, and messenger application of the user terminal through the communication unit, and processes the natural language, based on the reference text and the surrounding words of the reference text. The tool product purchase information and tool type information including tool type information and tool-specific dimension information according to tool product purchase are generated, and the purchase pattern and the detail are based on the tool product purchase information and the tool product detail information. Calculate the pattern,
The word surrounding the reference word and a predetermined threshold value based on a lexical dictionary including semantic lexical list relation information and providing a semantic relationship of a high word, a lower word, an equal word, and a whole word for a set of synonyms. Select the type of the tool and the reference of the tool from the surrounding words to generate the tool product purchase information and the tool product detail information,
The recommendation unit,
Types of specifications for each tool and specifications according to the specifications of the body diameter, the opposite side distance, the fit distance, the head height, the fillet radius and the tread length of the bolts and the specifications of the total length, the tread length and the tensile strength of the anchors Considering the different numerical values for each type of tool to build the non-supervised learning based on the recommendation model based on the dimensions of the tool according to the specifications for each type to generate the customized tool recommendation information according to the pattern for each dimension,
The recommendation unit predicts a purchase date of a new tool product purchase based on the purchase pattern, and generates the customized tool recommendation information including a type and a value of a tool predicted before the purchase date of the new tool product purchase. Apparatus for providing AI-based customized tool recommendation information to be transmitted to the user terminal. delete The method of claim 1,
The pattern calculation unit,
Artificial intelligence-based tailored to calculate the purchase pattern and the history pattern of the individual user by building a non-supervised learning-based pattern calculation model to input the tool product purchase information and the tool product details information of the individual user Tool recommendation information providing device. The method of claim 3,
The pattern calculation unit,
Constructing a non-supervised learning-based pattern calculation model for inputting the tool product purchase information and the tool product detail information of all the users to calculate the purchase pattern and the detail pattern of the all users;
The recommendation unit,
And generating customized tool recommendation information for the new tool product request based on the purchase pattern and the history pattern of all the users. delete delete The method of claim 1,
The pattern calculation unit,
Form nouns extracted from the text are transformed into vector spaces through morphological analysis.
AI-based customized tool recommendation information providing apparatus for selecting the surrounding words based on the distance between the vectors of the noun. The method of claim 3,
The type of tool is
Hex bolt, heavy hex screw, hex cap screw, U bolt, set anchor, all anchor, wing bolt, eye bolt, hinge bolt, hex nut, square nut, wing nut, eye nut, tab washer Will, AI-based customized tool recommendation information providing device. The method of claim 8,
The recommendation unit,
An apparatus for providing customized tool recommendation information based on artificial intelligence, which generates the customized tool recommendation information according to the pattern for each dimension by building a dimension recommendation model based on non-supervised learning based on the tool dimension information according to the tool type. . In the method of providing AI-based customized tool recommendation information,
Tool product purchase information and tool type information and tool-specific information including the purchase time information of the tool product for purchasing the tool product including nuts, bolts, anchors and washers used by the communication unit in the industrial field from the user terminal Receiving tool product detail information including information;
From the tool product purchase information and the tool product detail information, the pattern calculation unit includes a purchase pattern including a pattern for each month, week, and time, and a detail pattern including a pattern for each type of tool and a pattern for dimension considering different specifications according to shapes of tool types. Calculating;
Generating, by the recommendation unit, customized tool recommendation information including tool dimension information based on specifications for each type of customized tool for a new tool product purchase request based on the purchase pattern and the detail pattern; And
The communication unit transmitting the customized tool recommendation information to the user terminal,
The pattern calculation unit
Collect the text input through the E-mail, text message, MMS, messenger application of the user terminal through the communication unit to process the natural language, based on the reference text and the surrounding words of the reference text of the natural language processed text in the purchase of tool products Generating tool product detail information including the tool product purchase information and tool type information and tool-specific dimension information, and calculating the purchase pattern and the detail pattern based on the tool product purchase information and the tool product detail information,
The word surrounding the reference word and a predetermined threshold value based on a lexical dictionary including semantic lexical list relation information and providing a semantic relationship of a high word, a lower word, an equal word, and a whole word for a set of synonyms. Select the type of the tool and the reference of the tool from the surrounding words to generate the tool product purchase information and the tool product detail information;
The recommendation unit,
Types of specifications for each tool and specifications according to the specifications of the body diameter, the opposite side distance, the fit distance, the head height, the fillet radius and the tread length of the bolts and the specifications of the total length, the tread length and the tensile strength of the anchors Considering the different numerical values for each type of tool to build the non-supervised learning based on the recommendation model based on the dimensions of the tool according to the specifications for each type to generate the customized tool recommendation information according to the pattern for each dimension,
The recommendation unit predicts a purchase date of a new tool product purchase based on the purchase pattern, and generates the customized tool recommendation information including a type and a value of a tool predicted before the purchase date of the new tool product purchase. How to send to the user terminal, AI-based customized tool recommendation information providing method. delete The method of claim 10,
Computing the purchase pattern and the history pattern,
Calculate the purchase pattern and the detail pattern of the individual user by constructing a non-supervised learning-based pattern calculation model using the tool product purchase information and the tool product detail information of the individual user as inputs,
The purchase pattern includes a monthly, weekly, time-specific pattern, and the history pattern, a tool-specific pattern and a dimension-specific pattern, artificial intelligence based customized tool recommendation information providing method. The method of claim 12,
Computing the purchase pattern and the history pattern,
Constructing a non-supervised learning-based pattern calculation model for inputting the tool product purchase information and the tool product detail information of all the users to calculate the purchase pattern and the detail pattern of the all users;
Generating the customized tool recommendation information,
Method for providing customized tool recommendation information based on artificial intelligence that generates customized tool recommendation information for the new tool product request based on the purchase pattern and history pattern of the entire user. delete delete The method of claim 10,
Computing the purchase pattern and the history pattern,
Form nouns extracted from the text are transformed into vector spaces through morphological analysis.
AI-based customized tool recommendation information providing method for selecting the surrounding words based on the distance between the vectors of the noun. The method of claim 12,
The type of tool is
Hex bolt, heavy hex screw, hex cap screw, U bolt, set anchor, all anchor, wing bolt, eye bolt, hinge bolt, hex nut, square nut, wing nut, eye nut, tab washer How to provide AI-based customized tool recommendations. The method of claim 17,
Generating the customized tool recommendation information,
The method for providing customized tool recommendation information based on artificial intelligence is to generate the customized tool recommendation information based on the pattern for each dimension by building a dimension recommendation model based on non-supervised learning based on the tool dimension information according to the tool type. . 19. A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 10, 12, 13, and 16-18.

Images