KR102279210B1 - Electronic device for determining similarity between sequences considering item classification scheme and control method thereof - Google Patents

Abstract

An embodiment according to the present invention includes the steps of: obtaining a first sequence consisting of first items having an order; obtaining a second sequence consisting of second items having an order; and determining whether the first and second sequences are similar based on the first information and the second information, wherein the first information converts one of the first items into one of the second items wherein the second information comprises information about the length of a shortest path in the item classification tree between one of the first items and one of the second items. A control method for determining the similarity between sequences in consideration of the item classification system may be provided.

Description

Electronic device for determining similarity between sequences considering item classification scheme and control method thereof

The present invention relates to an electronic device for determining the degree of similarity between sequences in consideration of an item classification system and a control method therefor, and more particularly, to determine the similarity between sequences including items by using a distance of a shortest path between items in an item classification system It relates to an electronic device and a control method.

Due to the recent spread of smartphones and the development of social network services, a big data environment has arrived in which various data are continuously generated in the form of data streams from smartphone users in real time. The data stream refers to data that is sequentially and infinitely generated according to the passage of time.

Similar sequence matching is a problem of searching for a data sequence similar to a query sequence given by a user from among numerous data sequences. When similar sequence matching is performed using numerical data having continuity as data constituting a sequence, a similar data sequence having a similar shape can be found.

As an example of the data sequence, purchase history data is one of representative data generated by a customer. The purchase history data contains the characteristics of the person purchasing the product, the products consumed, and consumption patterns. By comparing the purchase history data of other customers and determining whether they are similar, a service such as a new product recommendation can be provided.

However, the conventional method for determining similarity between purchase history data does not reflect information on the correlation between products in a sequence, so there is a problem in that accuracy of similarity determination of purchase history data is lowered.

Korean Patent Publication No. 10-1937989 (Set-based Similar Sequence Matching Apparatus and Method)

An electronic device for determining the degree of similarity between sequences in consideration of an item classification system and a control method therefor according to an embodiment of the present invention aims to provide a degree of similarity between sequences composed of items having an order.

An electronic device for determining a degree of similarity between sequences in consideration of an item classification system and a control method therefor according to an embodiment of the present invention aims to provide a degree of similarity between purchase histories of products.

An electronic device for determining a degree of similarity between sequences in consideration of an item classification system and a control method therefor according to an embodiment of the present invention aims to use a product classification tree to provide a degree of similarity between purchase histories of products.

An electronic device for determining the degree of similarity between sequences in consideration of an item classification system and a control method therefor according to an embodiment of the present invention aims to shorten a time for determining the degree of similarity between purchase histories of products.

According to an embodiment of the present invention, a process of obtaining a first sequence consisting of first items having an order; obtaining a second sequence consisting of second items having an order; and determining whether the first and second sequences are similar based on the first information and the second information, wherein the first information converts one of the first items into one of the second items wherein the second information comprises information about the length of a shortest path in the item classification tree between one of the first items and one of the second items. A control method for determining the similarity between sequences in consideration of the item classification system may be provided.

An embodiment of the present invention may provide a control method for determining the degree of similarity between sequences in consideration of an item classification system, characterized in that the first and second items correspond to terminal nodes of the same item classification tree.

In an embodiment of the present invention, the second information further includes information on the length of the longest path of the items in the item classification tree. A control method for determining the similarity between sequences in consideration of the item classification system. can provide

According to an embodiment of the present invention, the length of the shortest path in the item classification tree between one of the first items and one of the second items is based on pre-stored depth information for each node of the item classification tree. It is possible to provide a control method for determining the degree of similarity between sequences in consideration of the item classification system, characterized in that it is determined by

An embodiment of the present invention, a memory for storing information about the item classification tree; and a processor for controlling the memory, wherein the processor obtains a first sequence consisting of first items having an order, obtains a second sequence consisting of second items having an order, and the first information and controlling to determine whether the first and second sequences are similar based on second information, wherein the first information is based on the number of operations for converting one of the first items into one of the second items. wherein the second information comprises information about a length of a shortest path in the item classification tree between one of the first items and one of the second items. An electronic device for determining the degree of similarity between each other may be provided.

An embodiment of the present invention may provide an electronic device for determining the degree of similarity between sequences in consideration of an item classification system, wherein the first and second items correspond to terminal nodes of the same item classification tree.

In an embodiment of the present invention, the second information further includes information on the length of the longest path of the items in the item classification tree. An electronic device for determining the similarity between sequences in consideration of the item classification system. can provide

In one embodiment of the present invention, the length of the shortest path in the item classification tree between one of the first items and one of the second items is a depth for each node of the item classification tree pre-stored in the memory. An electronic device for determining the degree of similarity between sequences in consideration of an item classification system, which is determined based on information, may be provided.

An embodiment of the present invention provides a control method for determining the degree of similarity between sequences in consideration of the item classification system, thereby providing a quantitative numerical value rather than providing discontinuous results such as similarity and dissimilarity for the similarity between sequences. It has the effect of providing similarity.

An embodiment of the present invention has an effect of comparing whether sequences having the same item classification system are similar.

An embodiment of the present invention provides a control method for determining the degree of similarity between sequences in consideration of an item classification system based on the previously stored depth information on each node of the item classification tree, thereby reducing the similarity determination time of sequences. have an effect

An embodiment of the present invention has an effect of providing a degree of similarity between product purchase histories by providing a control method for determining a degree of similarity between sequences in consideration of an item classification system.

1 shows a comparison between the Euclidean distance and the DTW distance.
Figure 2 (a) shows a part of the product classification tree used in the present invention.
Figure 2 (b) shows an example of the longestPath and itemPath of the product classification tree used in the present invention.
3 is a measurement result of the execution time applied to the proposed method (Experimental results for proposed methods)
4 is a graph comparing the performance of a simple technique and a proposed technique to which a segment tree is applied according to a purchase history length.
5 is a graph comparing the performance of the simple technique and the proposed technique according to the number of products in the product classification system tree.
6 is a flowchart illustrating a control method for determining a degree of similarity between sequences in consideration of an item classification system according to an embodiment of the present invention.
7 and 8 are block diagrams of an electronic device 1000 according to an exemplary embodiment.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed in this specification are only exemplified for the purpose of explaining the embodiments according to the concept of the present invention, and the embodiments according to the concept of the present invention are It may be implemented in various forms and is not limited to the embodiments described herein.

Terms such as first or second may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another, for example without departing from the scope of the inventive concept, a first component may be termed a second component and similarly a second component A component may also be referred to as a first component.

The technical terms used herein are used only to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present specification, terms such as "comprise" or "have" are intended to designate that a feature, number, step, operation, component, part, or combination thereof described herein exists, but one or more other features It is to be understood that it does not preclude the possibility of the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, are common to those of ordinary skill in the art to which the technology disclosed herein belongs, unless specifically defined otherwise herein. has the same meaning as understood as Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present specification. shouldn't

Hereinafter, a machine learning model (method) of an electronic device for determining the degree of similarity between sequences in consideration of an item classification system according to an embodiment of the present invention and a control method thereof will be described with reference to the accompanying drawings.

Sequence data of items for items having a tree-type item classification system means a set of items having an order. As a specific example of this, for products having a tree-type product classification system, the product purchase history is sequence data in which products according to the purchase order are listed. The following description of the purchase history of the product may be directly applied to sequence data for items having an order.

1. Introduction

With the advent of the big data era, customers are generating huge amounts of new digital data on their own. The generated data becomes the characteristics of the customer itself, and furthermore, the customer can be expressed as data. The purchase history data is one of the representative data generated by the customer. The purchase history data contains the characteristics of the person purchasing the product, the products consumed, and consumption patterns. Therefore, a company needs the ability to analyze given customer data, and it can be reflected in decision making based on the analyzed results.

The present invention proposes a new similarity method for calculating how similar the purchase histories of any two customers are in the purchase history data composed of purchase items of customers over time. The similarity measurement method proposed in the present invention is extended to reflect the hierarchical relationship between items as well as the existing similarity method for measuring the similarity between data in which the order of items exists, such as purchase history data.

The purchase history data is one of the representative sequence data as data containing the order between purchase items. A sequence means data in which an order exists between items of data, and even if two sequences have the same configuration, if the order is different, the two sequences are different sequences. Examples of the sequence may include not only purchase history data, but also web logs, protein sequences, and the like.

The similarity measurement method proposed in the present invention not only calculates the similarity in consideration of the order between items in a sequence, unlike existing purchase history data analysis methods, but also considers different similarities between purchased products using a product classification system.

In the case of measuring the similarity between purchase history sequences using only existing methods, only the consistency of products constituting the sequence is the criterion for determining the similarity. Sequence 1 is an example of a purchase history sequence.

[Sequence 1]

,

의 유사도와

,

의 유사도를 동일하게 계산한다. 시퀀스 1의 시퀀스들은 모두 두 번째 구매 상품이 동일하기 때문에 각 시퀀스의 첫 번째 구매 상품만이 시퀀스 간 유사도에 영향을 미친다. For example, if the purchase history sequence is the same as sequence 1, the existing similarity measurement method is the sequence

,

with the similarity of

,

Calculate the similarity of Since all the sequences of Sequence 1 have the same second purchase product, only the first purchased product in each sequence affects the degree of similarity between the sequences.

의 'Coke'와

의 'Sprite'를 비교하는 경우와

의 'Coke'와

의 'Jeans'를 비교하는 경우 같은 유사도를 부여한다. Because the existing methods only judge whether the products match or not,

'Coke' of

When comparing the 'Sprite' of

'Coke' of

When comparing 'Jeans' of 'Jeans', the same degree of similarity is given.

However, in reality, 'Sprite' belongs to the same category of 'Coke' and beverages relative to 'Jeans'. Considering this, it is necessary to give a higher similarity than 'Jeans' when comparing 'Coke' and 'Sprite'. Therefore, in the present invention, the similarity between two purchase histories is more accurately calculated by extending the existing sequence similarity measurement method to consider the product classification system when calculating the similarity between purchase history sequences.

In addition, in the present invention, since the similarity is calculated in consideration of the order of items, the Levenshtein distance and dynamic time warping (DTW), which are representative sequence similarity measurement methods, are representative in order to select a similarity measurement method between sequences that affects the similarity calculation performance. Dynamic Time Warping) distance and the performance of Needleman-Wunsch similarity were compared.

Through an experiment using virtual sequence data, it was confirmed that the dynamic time warping distance, which not only considers the product classification system, but also accurately measures the similarity for two sequences having different lengths, is the most suitable similarity measurement method.

The structure of the specification of the present invention is as follows. Chapter 2 examines the sequence similarity measurement methods used in the present invention. In Chapter 3, the similarity measurement method proposed in the present invention will be described in detail. Chapter 4 shows the performance evaluation results of the proposed method and the existing method, and Chapter 5 concludes.

2. Related Studies

2.1 Analysis of purchase history data

The large-scale purchase history data containing the consumer's consumption activity contains the customer's consumption pattern. The process of finding a purchase pattern that occurs frequently in the customer's purchase history is called correlation analysis. Association analysis creates rules by judging each product as an independent entity, where shopping cart analysis considers even the hierarchical relationship between items. Contrary to this, sequential pattern analysis creates rules in consideration of the precedence and precedence of purchase details. All three analyzes generate rules between items using evaluation criteria of support, confidence, and lift.

A typical example of using customer purchase history data is Wal-Mart in the United States. Wal-Mart was the first company to pay attention to customer shopping carts, which directly affect sales, and analyzed vast amounts of data to uncover the relationship between beer and diapers. Companies that engage in various marketing activities to increase sales can use this data to identify the actual relationship between products and use it as a marketing strategy.

2.2 How to Measure Sequence Similarity

A sequence is data made up of two or more items, and it is data that contains an order between these items. Representative sequence data includes web log data and protein sequence data. You can find protein sequences with similar functions by grouping similar users in web log files by analyzing sequence data or by grouping protein sequences with similar structures. It is important to define the degree of similarity by considering the order in sequence data in which there is a precedence relationship between items, and it can be divided as follows according to the method of calculating the degree of similarity.

1) Edit-based similarity measurement method

This is an algorithm to find the minimum number of correction operations to make two strings the same. The correction operation refers to add, substitute, and delete operations, and the most representative algorithm is the Levenshtein distance algorithm. The number of correction operations is used as a measure of similarity, and the smaller the value, the more similar the two strings are judged.

2) How to measure alignment similarity

It is mainly used to find similar regions between two sequences when analyzing the correlation between protein sequences or nucleic acid sequences. According to the sort range, there are local sorting methods and global sorting methods. A representative local sorting algorithm is the Smith-Waterman algorithm, and the global sorting algorithm is Needleman-Wunsch (Needleman-Wunsch). ) is an algorithm. Both algorithms sort using gaps so that the two sequences are most similar according to the sort range.

3) Set-based similarity measurement method

It is calculated by converting a string into a set of characters or a set of tokens. It operates using a set relationship, and when a string is divided into tokens, the string is cut into N standard units of length using the N-gram concept. A typical algorithm is Jaccard similarity. Jacquard similarity indicates the ratio between union and intersection between elements constituting a set, and has a value between 0 and 1.

2.3 Method of measuring similarity of purchase history sequence

In this section, the Levenstein distance, dynamic time warping similarity, and Needlemann-Bnish similarity used for performance comparison when measuring sequence similarity proposed in the present invention will be described in detail.

1) Levenshtein street

The Levenstein distance, also known as the edit distance algorithm, refers to the minimum number of operations required to convert one string to another. Modification operations between two strings are insert, substitute, and delete operations. The characters of the two strings to be compared are compared one character at a time. For addition and deletion, the operation cost is given as 1, and for the replacement operation, 0 or 1 is assigned depending on whether the characters match and the similarity is calculated.

2) Dynamic Time Warping (DTW) Distance

It is an algorithm that measures the similarity of two time series patterns with different speeds, and is used for speech recognition and handwritten character recognition. Since the distance is calculated while moving in the direction that minimizes the distance between the two time series, it can be calculated even for a partially distorted or deformed waveform, unlike when calculating with the Euclidean distance.

1 shows a comparison between the Euclidean distance and the DTW distance.

벡터로 표현된다고 하자. 도 1은 두 시계열 그래프 A, B에 대하여 유클리디안(Euclidean) 유사도로 계산한 경우와 동적 타임 워핑 유사도로 계산한 경우이다. 동적 타임 워핑 유사도로 계산한 경우는 유클리디안 유사도와 다르게 시계열 그래프의 한 점에서 다른 시계열 그래프의 하나 혹은 그 이상의 점에 대응하여 계산 할 수 있다. 이러한 특성으로 인해 서로 다른 길이의 시퀀스에 대해서도 효과적으로 유사도를 계산할 수 있다. For example, if two time series graphs A and B are

Let it be expressed as a vector. 1 is a case in which two time series graphs A and B are calculated with Euclidean similarity and dynamic time warping similarity. In the case of calculating the dynamic time warping similarity, it can be calculated in correspondence with one or more points in the other time series graph from one point in the time series graph, unlike the Euclidean similarity. Due to these characteristics, similarity can be effectively calculated even for sequences of different lengths.

3) Needleman-Wunsch similarity

As an algorithm for sequence comparison between sequences of proteins or nucleotides in the field of bioinformatics, a gap is used to align two sequences so that the two sequences have the highest similarity. The gap penalty, match, and mismatch values are user-specified values, and the alignment results between sequences vary depending on the values. The similarity between two sequences is calculated using the cost of insertion, deletion, and matching operations for two sequences, such as the Levenstein distance.

In the present invention, the similarity was measured using the three similarity measurement methods. All of them can compare items between two sequences one by one and do not require separate operations such as dividing the sequence into set elements. However, in the case of comparing purchase history sequences using only the existing method, since similarity is calculated by assigning only a value of 0 or 1 by judging only whether products match or not, in the present invention, this is extended to consider the hierarchical relationship between products.

The proposed method of the present invention uses a product classification tree to assign a similarity between products as a value between 0 and 1 to further subdivide the degree of association between each product when calculating the similarity. Therefore, in the present invention, a part of the process was modified to reflect the product classification system when calculating the similarity between purchase history sequences, and Section 4.2 shows the performance evaluation results for them.

3. Measurement of sequence similarity considering the product classification system

In this chapter, the proposed method for calculating the similarity between two sequences in the purchase history data using the product classification system proposed in the present invention will be described in detail.

3.1 Overview

The similarity measurement method between sequences proposed in the present invention measures the similarity between two sequences by using a product classification system given in purchase history data. In the purchase history data, one purchase history is sequence data in which items sequentially purchased by one customer are listed. When these two arbitrary sequence data are given, the similarity measuring method proposed in the present invention measures the similarity in consideration of the order between the items. In addition, it is different from the existing sequence similarity measurement method in that the similarity is calculated by considering the classification system between products using a given product classification system.

The similarity measurement method between purchase history sequences proposed in the present invention calculates the similarity between items in a sequence using the Levenstein distance, the dynamic time warping distance, and the Needleman-Bnish similarity, which are representative sequence similarity measurement methods. In order to reflect the product classification system when calculating the similarity, a part of the execution process was modified to subdivide the correlation between the two items to be compared.

3.2 Problem Definition

In this section, the concept of a product classification tree used in calculating sequence similarity in the present invention and similar purchase history defined in the present invention will be described.

1) Product classification tree

Figure 2 (a) shows a part of the product classification tree used in the present invention. Figure 2 (b) shows an example of the longestPath and itemPath of the product classification tree used in the present invention.

Referring to FIG. 2( a ), in general, distribution companies such as department stores and marts have a product classification system in which layers are divided such as large/medium/small categories for all products. For example, a product called 'pants' first belongs to the subcategory 'bottoms', and here the bottoms again belong to the middle category 'clothes'. In this way, all products occupy the lowest part of the product classification system. In the present invention, such a product classification system is expressed as a tree data structure.

The system of the product classification tree used in the present invention was constructed with reference to the classification system of Amazon, an American electronic commerce company. In each node of the tree, the name of the actual product or the upper category of the product is stored.

In the product classification tree of Fig. 2(a), the leaf node of the tree corresponds to the product (eg, a, b, c, d, e, f, g, h, i, j, k), and the inside The internal node is the parent category for each product (eg C1, C2, C3, C4, C5, C6, C7, C8). In addition, the top node in the product classification tree is expressed as a root.

This product classification tree is used to compare similarities between purchase history sequences. Using the product classification tree, the items in the sequence are quantified according to different degrees of association in the product classification tree during the algorithm execution process and reflected in the calculation.

In the purchase history data, all items in the sequence correspond to leaf nodes of the product classification tree, and the terminal nodes correspond to products.

The similarity measurement method between purchase history sequences proposed in the present invention calculates the similarity between items using the product classification system shown in FIG. 2 . Therefore, even if the two products do not match, the degree of similarity between the products can be subdivided and calculated because the calculation reflects the degree of closeness between the two products in the product classification tree.

2) Definition of similar purchase history

개의 구매 항목으로 이루어진 시퀀스

를

라 하자.

는 시퀀스

의

번째 구매 항목을 나타내며, 이 시퀀스는 상품

부터

까지 순서대로 구매한 이력 데이터를 나타낸다. |S|는 시퀀스의 크기 또는 시퀀스 내 구매된 항목의 개수를 나타낸다.The purchase history data calculated in the present invention is sequence data in which items purchased by each customer are listed in order. In the purchase history data, each sequence represents the purchase history of one customer,

sequence of purchases

to

let's say

is the sequence

of

Represents the second purchase, this sequence is

from

Shows purchase history data in order up to. |S| indicates the size of the sequence or the number of purchased items in the sequence.

In the method for measuring the similarity between sequences proposed in the present invention, even if detailed items of two sequences are different, it is determined that the items are similar when the upper category of the items is the same.

Sequence 2 is an example of a similar purchase history defined in the present invention.

[Sequence 2]

,

는 두 고객에 대한 구매이력 데이터로,

과

는 모두 서로 다른 구매항목으로 구성되어 있다. 하지만 두 시퀀스의 세부 항목들은 매우 연관성이 높으며, 유사한 구매 순서를 보인다. 시퀀스 내 각 항목을 상위 범주로 바꿀 경우, 두 시퀀스 모두 음료, 간식, 의류 범주에 속한 상품 순으로 구성된 구매 이력인 것을 볼 수 있다. 이처럼 두 상품이 완전히 동일하지 않더라도 동일한 상위 범주에 속한다면 그렇지 않은 경우보다 더 높은 유사도를 부여할 필요가 있다. 따라서 본 발명에서는 상품 분류 트리를 활용하여 두 항목들이 다르더라도 서로 동일한 상위 범주를 가진다면 더 높은 유사도를 부여하여 계산한다. two sequences in sequence 2

,

is the purchase history data for two customers,

and

are all made up of different purchase items. However, the details of the two sequences are highly related and show a similar purchase order. If you change each item in the sequence to a higher category, you can see that both sequences are purchase histories consisting of products belonging to the beverage, snack, and clothing categories in order. Even if two products are not completely identical, if they belong to the same upper category, it is necessary to give a higher degree of similarity than if they were not. Therefore, in the present invention, using the product classification tree, even if two items are different, if they have the same upper category, a higher degree of similarity is given and calculated.

3.3 Measuring Sequence Similarity

This section describes in detail the similarity measurement method for purchase history sequences using the product classification system using the Levenstein distance, dynamic time warping distance, and Needlemann-Bnish similarity, which are the representative methods for measuring similarity between sequences described in Section 2.3. .

In order to apply the product classification tree, a part of the algorithm's execution process has been modified, and the square part of the pseudocode of codes 1, 2, and 3 below is a newly modified part to apply the proposed method.

과

에 대한 2차원 배열

을 생성한다. 각 알고리즘의 배열

의 값은 이전 원소(

,

,

)를 활용하여 구한다(

,

). 배열

의 마지막 원소가 각 알고리즘 별 유사도 값이 되며, 알고리즘 별로 다른 유사도 값의 범위를 가진다. 다음은 각 알고리즘 별 유사도 측정 방법에 대하여 상세히 설명한다.All three algorithms use the purchase history sequence to calculate the similarity.

and

2D array for

create array of each algorithm

The value of the previous element (

,

,

) to find (

,

). Arrangement

The last element of is the similarity value for each algorithm, and each algorithm has a different similarity value range. The following describes in detail the similarity measurement method for each algorithm.

1) Levenshtein street

,

)는 0에서

사이 값을 가진다. The Levenstein distance compares two strings one character at a time, and after calculation, the value of the last element of the array becomes the minimum editing distance of the two strings. Levenshtein Dist(

,

) is from 0

has a value between

,

에 대하여 문자열이 없는 경우도 포함하여 2차원 배열

을 행의 개수가

이고 열의 개수가

인

크기로 생성한다. 배열의 첫 번째 행과 열은 0에서부터 두 문자열의 길이만큼 증가시켜가며 초기화한다. 배열의 두 번째 행과 열부터는 이전까지 계산된 배열 값을 사용하여 채워나간다.two sequences

,

2D array including no string for

is the number of rows

and the number of columns is

sign

create in size The first row and column of the array are initialized by increasing the length from 0 to the length of the two strings. From the second row and column of the array, the array values calculated previously are used to fill in.

Code 1 relates to the pseudo-code of Levenshtein distance algorithm.

<Code 1>

에서 행에 해당하는

은 원본 문자열을 의미하며 열에 해당하는

는 바꾸고자 하는 목적 문자열을 의미한다.

는 추가(

), 대체(

cost), 삭제(

) 비용 중 가장 작은 값으로 채워진다. 삭제와 추가 연산은 이전 배열 값에 삭제, 추가 비용 1을 더한다. 대체 연산은 만일 현재 계산하는

번째 문자와

번째 문자가 일치한다면 대체 비용(cost)은 0을 더해주고, 다르다면 1을 더해준다. 코드 1은 레벤슈타인 거리 알고리즘을 활용하여 본 발명에서 제안하는 구매이력 시퀀스 유사도 계산 방법을 나타내는 의사코드이다. 코드 1의 의사코드 중 네모 칸은 새롭게 제안한 대체 연산 비용을 계산하는 함수로 코드 2의 의사코드가 수행된다. 만일 해당 부분이 문자 일치 유무에 따라 0 또는 1의 값을 가진다면 기존 레벤슈타인 거리 알고리즘이 수행된다.

corresponding to the row in

stands for the original string and corresponds to the column

means the target string to be changed.

is added (

), how(

cost), delete (

) is filled with the smallest of the costs. Delete and append operations add 1 delete and add cost to the previous array value. The replacement operation is if the current computation

second letter and

If the first character matches, the replacement cost adds 0, otherwise adds 1. Code 1 is a pseudo-code indicating a method for calculating the similarity of a purchase history sequence proposed in the present invention by using the Levenstein distance algorithm. Among the pseudocodes of code 1, the square box is a function that calculates the newly proposed replacement operation cost, and the pseudocode of code 2 is executed. If the corresponding part has a value of 0 or 1 depending on whether characters match or not, the existing Levenstein distance algorithm is performed.

For example, assuming that there are two sequences S1 = [a, e, m], S2 = [a, d, c] having different character strings, the operation required to convert the character string of S1 to the character string of S2 is Find the minimum number of times. dist(i, j) represents the editing distance of the character string up to the i-th of S1 and the character string up to the S2-th.

dist(i, j) = min [dist(i-1, j-1)+c(i, j), dist(i-1, j)+1, dist(i, j-1)+1] and It is defined as c(i, j) = 0 (if xi=yj) or 1 (otherwise). The edit distance of each character string of the sequences S1 and S2 is shown in Table 1 below.

[Table 1]

The algorithm proposed in the present invention calculates the similarity between products more accurately by subdividing the value of the added cost during the replacement operation of the existing Levenstein distance algorithm from 0 to 1 using the product classification tree. Therefore, the substitution calculation cost proposed in the present invention is determined to be completely unrelated only in the case of the two most distant items in the product classification tree, and a maximum value of 1 is given. Otherwise, the shortest distance between two different products in the product classification tree. Subdivide the value of the replacement computation cost to have the length of the path. Code 2 is a pseudo-code for a proposed method of calculating the degree of association between two products by subdividing the degree of association between two products by using a product classification tree, unlike the existing alternative computation cost calculation method.

Code 2 relates to the pseudo-code of the proposed method.

<Code 2>

If the two products in the sequence to be compared are different, the product classification tree is searched to calculate the degree of association between the two products. Equation 1 is an equation for calculating the correlation between two product categories by using the product classification tree proposed in the present invention.

[Equation 1]

번째 항목과

번째 항목과의 가까운 정도이다. cost 값은 상품 분류 트리 내에서 찾고자 하는 두 항목의 최단 경로의 길이가 가장 먼 두 항목의 길이에서 차지하는 비율을 나타낸다. cost in the product classification tree

second item and

It is close to the second item. The cost value represents the ratio of the length of the shortest path of the two items to be found in the product classification tree to the length of the two items that are farthest.

longestPath is the length of the path of the two most distant nodes in the product classification tree, that is, the number of edges of the two most unrelated items in the product classification tree.

번째 항목과

번째 항목과의 최단 경로의 길이이다. 두 항목 사이의 최단 경로는 두 항목에서 가장 가까운 범주까지의 엣지(edge)의 수이다. 따라서 itemPath가 작을수록 두 항목은 높은 연관성을 가지며 가까운 범주에 분류되어 있는 상품임을 의미한다. itemPath is within the product classification tree.

second item and

It is the length of the shortest path to the second item. The shortest path between two items is the number of edges from the two items to the nearest category. Therefore, the smaller the itemPath, the higher the correlation between the two items and the product is classified into a close category.

If itemPath is 0, both products are the same node and cost is 0. If itemPath is 1, it has the same value as longestPath, meaning the two items with the lowest correlation among the items in the tree, and the cost becomes 1.

In the case of the Levenstein distance algorithm, the newly calculated cost is a cost added during the replacement operation and is used to select the minimum value among the addition, replacement, and deletion costs.

For example, New c(i, j) is substituted for c(i, j) = 0 (if xi=yj) or 1 (otherwise) described above, and New c(i, j) is Equation 1 It can be expressed as the cost of

Referring to FIG. 2(b), Candy and Polos are the two items with the lowest correlation, and lognestPath = 7, and it can be seen that the shortest path of Coke and Sprite in Beverage, which is the same node (C5), is itemPath = 2. have.

2) Dynamic Time Warping (DTW) Distance

,

에 대하여

크기의 2차원 무한대 값을 가지는 배열

을 생성한다.

는 레벤슈타인 거리 알고리즘과 달리 비교하고자 하는 두 항목에 대한 비용(cost)을 먼저 계산한 후, 삭제(

), 일치(

), 삽입(

) 비용 중 최솟값에 더한다. The dynamic time warping algorithm calculates the distance using the accumulated distance by matching while moving in a direction that minimizes the distance between the two time series. sequence

,

about

an array of two-dimensional infinity values

create

is different from the Levenstein distance algorithm, calculates the cost of the two items to be compared first, and then deletes (

), Same(

), insert (

) to the minimum of the costs.

)인 경우 기존의 동적 타임 워핑 알고리즘이 수행된다. 본 발명에서는

의 한 지점에서

의 다른 지점으로 할당되는 매핑(mapping) 비용을 두 값의 차이가 아닌 제안 방법을 통하여 0에서 1 사이의 값을 가지도록 하였다.Among the pseudocodes of Code 3, the squares are pseudocodes for the dynamic time warping similarity to which the proposed method of Code 2 is applied, and the corresponding part is the difference between the two values (

), the existing dynamic time warping algorithm is performed. In the present invention

at a point in

The mapping cost, which is allocated to different points of , is set to have a value between 0 and 1 through the proposed method, rather than the difference between the two values.

Code 3 relates to the pseudo-code of DTW algorithm.

<Code 3>

3) Needleman-Wunsch (NW: Needleman-Wunsch) similarity

,

에 대하여

크기의 2차원 배열

을 생성하여 0으로 초기화한다. 알고리즘 수행을 위해 공백 페널티(gap penalty), 일치 보상(match award), 불일치 페널티(mismatch penalty) 세 변수에 대한 사용자 정의 값을 지정한다. 사용자 정의 값에 따라서 계산된 유사도 값의 범위가 달라지며, 본 발명에서는 상품 분류 트리를 활용한 연산 비용은 0에서 1 사이의 값을 가지므로 일치 보상은 1, 불일치 페널티는 0, 공백 페널티는 -1로 주었다. 사용자 정의 값이 다음과 같은 경우 니들만-브니쉬 유사도 NW Sim(

,

)는

사이 값을 가진다. The Needleman-Bnish algorithm aligns the two sequences with the highest degree of similarity using spaces while minimizing mutations between the two sequences. sequence

,

about

two-dimensional array of sizes

is created and initialized to 0. Specify user-defined values for three variables: gap penalty, match award, and mismatch penalty to perform the algorithm. The range of the calculated similarity value varies according to the user-defined value, and in the present invention, the calculation cost using the product classification tree has a value between 0 and 1, so the match reward is 1, the mismatch penalty is 0, and the blank penalty is - given as 1. Needleman-Varnish Similarity NW Sim(

,

) is

has a value between

Code 4 relates to the Pseudo-code of Needleman-Wunsch algorithm.

<Code 4>

,

)의 일치 유무에 따라서 일치 보상(match award) 혹은 불일치 페널티(mismatch penalty)가 부여된다면 기존의 니들만-브니쉬 거리가 계산된다.Code 4 is the pseudo code for the Needlemann-Bnish algorithm to which the proposed method is applied, and the square box is the pseudo code of code 2 for the proposed method for calculating the association between two products in the product classification tree. If the part wants to compare two items (

,

), if a match award or a mismatch penalty is given depending on the presence or absence of a match, the existing Needleman-Bnish distance is calculated.

4. Experimental results

This chapter shows the performance measurement results for the similarity measurement method between purchase history sequences using the product classification system proposed in the present invention. The comparison results of similarity measurements between the existing method and the proposed method with respect to the Levenstein distance, dynamic time warping distance, and Needlemann-Bnish similarity, which are representative similarity measurement methods described in Section 2.3, and the execution time comparison results for the three algorithms are shown.

4.1 Experimental environment and methods

The similarity measurement method between purchase history data sequences proposed in the present invention was implemented by dynamic programming using Python 3.7, and the product classification tree was implemented using the anytree library. The experiment was performed on a PC running Windows 10 operating system equipped with Intel i7-5820 3.3 GHz CPU and 8GB memory.

The product classification tree used in the experiment has a height of 5 and has a total of 37 nodes. The tree consists of 26 leaf nodes representing products and 11 nodes corresponding to product categories. In the experiment, virtual sequence data in which the number of items in the sequence ranges from 3 to 10 was generated and used.

Table 2 relates to Evaluation results for different similarity measures.

[Table 2]

4.2 Experimental results

This section shows the performance measurement results according to the similarity measurement method between various sequences.

1) Accuracy comparison experiment

First, it was evaluated whether the proposed method can more accurately calculate the similarity between sequences in the purchase history data compared to the existing sequence similarity measurement method.

Sequence 3 is virtual sequence data used in the accuracy measurement experiment. In Sequence 3, Example 1 and Example 2 are experimental data to measure whether the proposed method more accurately calculates the similarity between purchase history sequences compared to the existing similarity method. This is experimental data on purchase history. Finally, Example 3 is similarity measurement virtual sequence data for sequences having different sizes.

Sequence 3 relates to Examples of sequence data.

[Sequence 3]

,

과 예제 2의

,

는 서로 유사하지 않은 구매이력 시퀀스이며 예제 1의

,

과 예제 2의

,

는 서로 매우 유사도가 높은 구매 상품으로 구성된 시퀀스이다. 이를 통하여 제안 방법과 기존 방법과의 정확도를 비교하였다. 예제 3의 경우

는

과 시퀀스의 크기가 다르지만 크게 보면 동일한 범주의 구매순서(간식, 전자 제품, 음료 순)를 가진다. 이를 통해 서로 다른 길이를 가지는 시퀀스에 대해서도 정확히 유사도를 계산하는지 측정하였다.Example 1 of

,

and in Example 2

,

is a sequence of purchase history that are not similar to each other, and

,

and in Example 2

,

is a sequence composed of purchased products that are very similar to each other. Through this, the accuracy of the proposed method and the existing method was compared. For example 3

is

Although the size of the sequence is different, they have the same purchase order (snacks, electronic products, beverages) in the same category. Through this, it was measured whether similarity was accurately calculated even for sequences having different lengths.

)를 사용하여 초기화하기 때문에

,

의 거리와

,

의 거리가 다르게 계산되지만 상품 간의 연관 정도는 고려하지 않고 계산한다. 반면 제안 방법을 적용한 세 알고리즘의 경우 모두 상품 간의 유사도가 낮은

,

에 비하여 상대적으로 시퀀스 내 상품 간의 연관성이 높은

,

에 대하여 더 정확히 계산하는 것을 볼 수 있다. <Table 2> is a result table for the similarity measurement results in the three existing algorithms using the virtual sequence data of Sequence 3 and the similarity measurement results by applying the proposed method. As a result of the experiment for Example 1, in the case of the existing Levenstein distance and Needleman-Bnish similarity, the same value is calculated because the purchased products constituting the three sequences are all different. However, in the case of dynamic time warping distance, the difference (

) because it is initialized using

,

with the distance of

,

Although the distance of is calculated differently, it is calculated without considering the degree of association between products. On the other hand, in all three algorithms to which the proposed method is applied, the similarity between products is low.

,

relatively high correlation between products in the sequence compared to

,

It can be seen that a more accurate calculation of

Also, in the case of Example 2 using real data, it was confirmed through experiment that two sequences with higher correlation between products as in Example 1 showed a higher similarity than the case where they were not. Through this, the existing methods calculate by simply assigning 1 to different products, but when the proposed method is applied, the similarity is calculated more accurately even in a sequence composed of different products because the calculation is performed in consideration of the product classification tree.

과

의 길이가 다르더라도 더 정확하게 유사도가 계산된 것이라고 판단된다.As a result of the experiment in Example 3, in the case of Levenstein distance and Needlemann-Bnish similarity, both the proposed method and the existing method show the same similarity value even if they show a similar purchase order because only each product of the two sequences is compared. However, in the case of the dynamic time warping distance to which the proposed method is applied, a lower value (0.75) is calculated unlike the distance (3) measured by the existing method even if the length is different. This is because in the case of dynamic time warping distance, one item in a sequence is calculated by using the cumulative distance to correspond to multiple items in another sequence.

and

Even if the lengths of are different, it is judged that the similarity is calculated more accurately.

As a result of the experiment on the purchase history sequence, it was confirmed that all three algorithms to which the proposed method was applied measure the similarity more accurately than the existing algorithms because they consider the relevance between products. In addition, it was confirmed through experiments that the dynamic time warping algorithm accurately measures the similarity of sequences of different lengths, unlike the other two similarity measurement methods.

2) Performance speed comparison experiment

Next, the execution time of the proposed sequence similarity measurement technique was measured to evaluate whether it was an acceptable level. Experiments were performed on the Levenstein distance, the dynamic time warping distance, and the Needlemann-Bnish similarity to which the existing method and the proposed method were applied. When two sequences were given, the execution time was measured by increasing the number of similarity measurements between the sequences from 2000 to 10000 times. When measuring sequence similarity, virtual sequence data was generated to have a different random product composition and length each time.

3 is a measurement result of the execution time applied to the proposed method (Experimental results for proposed methods)

<Table 3> is a table showing the execution time of the existing method when the number of sequence similarity measurements is 10000 times. Referring to FIG. 3 , as a result of the experiment, the calculation speed of all three similarity comparison methods showed a tendency to depend on the number of products in the sequence and the composition of products in the sequence, and the execution time of the existing method and the proposed method was the median value after 10 measurements. took Table 3 relates to Evaluation results for conventional methods.

[Table 3]

As a result of the measurement, the execution time increases linearly as the number of comparisons increases, and all three algorithms show similar execution times. It is judged that all three algorithms show similar execution times because the algorithm's execution process is calculated using the two-dimensional array for the two sequences. In addition, compared to the execution time of the existing method in <Table 3>, the execution time of the proposed method is longer due to the product classification tree search process. However, the execution time of all three algorithms to which the proposed method is applied does not exceed a maximum of 2 minutes, indicating that the execution time of the proposed method is at a practically usable level.

5. Conclusion

In the present invention, by extending the existing sequence similarity measurement method, a new method for calculating the similarity between two sequences in consideration of the product classification system in the purchase history data is proposed. The purchase history sequence to be calculated in the present invention is data in which items purchased by one customer are listed in order. Given these two purchase history sequences and the product classification system, the existing similarity measurement method calculates the similarity by considering only the order of purchased products constituting the sequence. In this case, if the two sequences are purchase histories, the correlation between the two products to be compared is ignored. Therefore, in the present invention, not only the order of the constituent products in the sequence is considered, but also the similarity is more accurately calculated for the sequence composed of different products by using the product classification system.

In the present invention, in order to find a similarity measurement method suitable for a purchase history sequence, various sequence similarity measurement methods are considered. To this end, we used Levenstein distance, dynamic time warping distance, and Needlemann-Bnish similarity, which are currently representative methods for measuring sequence similarity, and some of these performance processes were modified to apply the proposed method. If the existing three algorithms simply assign 0 or 1 depending on whether or not products match when calculating the calculation cost between items in a sequence, the method proposed in the present invention uses a product classification tree to set the calculation cost to have a value between 0 and 1. segmented. The newly calculated computational cost means the path from the product classification tree of the two products to be compared to the closest common category, and 1 is assigned only when the closest common category is the root node. Through this, it was confirmed through experiment that different importance between products can be considered and more meaningful results are derived when calculated as independent entities. As a result of the accuracy measurement comparison experiment for the three algorithms, compared to the two similarity measurement methods with different dynamic time warping similarity, not only the degree of correlation of products in a sequence was considered, but also good performance was shown even when the lengths of the two sequences were different. It was confirmed that it is the most suitable measurement method when comparing similarity.

In the present invention, a process of searching a given product classification tree is performed in order to calculate the degree of association between two products even for different products during the sequence similarity measurement process. In future research, we plan to study a method for speeding up the product classification tree search for two sequences with various sequence lengths while increasing the product composition complexity in the sequence and a method for improving the similarity measurement performance.

Hereinafter, an improved part in the machine learning model (method) of an electronic device for determining the degree of similarity between sequences in consideration of the item classification system according to an embodiment of the present invention and a control method thereof will be described. The content of the present invention described above and the content of the present invention to be described later have a single technical idea, and the present invention can be configured by a combination of each part.

1. Introduction

Products purchased by each customer form a purchase history of that customer over time. Here, data in which an order exists between items is called a sequence. A company can analyze the purchase history or purchase sequence containing the characteristics and consumption patterns of customers and use it as a marketing strategy to increase sales. On the other hand, most products traded in the distribution industry have a classification system such as small classification, medium classification, and large classification. In this case, it can be said that the similarity between the two products is higher as the detailed classification is the same even if they are different products. For example, if two products belong to the same sub-category, the similarity is higher than if the two products belong to the same sub-category although they are in different sub-categories. So far, many studies have been conducted on methods for measuring the similarity between sequences. However, all of these studies only consider whether each item constituting a sequence is the same or different, and there have been few studies considering the classification system in the items.

Therefore, the present invention proposes a method for measuring similarity between purchase histories that considers not only the order of purchase of products but also the classification system existing in products. The method for measuring the similarity between purchase histories proposed in the present invention determines that the two purchase histories are similar even if the products of the two purchase histories are different, if the classification of the products is similar. The similarity measurement method proposed in the present invention extends the dynamic time warping (DTW) distance, which is a representative sequence similarity measurement method, to reflect the similarity between products according to the product classification system in the distance calculation. In addition, the present invention proposes an efficient calculation technique for the proposed similarity measurement method. The proposed method calculates the similarity between two products very quickly within the product classification system using a segment tree. Through this, the proposed method can be used very efficiently when it is necessary to compare multiple purchase histories.

The structure of the specification of the present invention is as follows. In Chapter 2, we briefly review the related research, and in Chapter 3, we describe the proposed method in detail. Section 4 shows the performance evaluation results of the proposed method, and Section 5 draws a conclusion.

2. Related Studies

A lot of research has already been done on how to measure the similarity between sequences. Among them, the dynamic time warping similarity is a method proposed to measure the similarity between two time series patterns with different speeds. The similarity is calculated using the accumulated distance while moving items in a direction that minimizes the distance between the two time series. It is known to calculate similarity very efficiently for sequences of different lengths.

In the present invention, several previous experiments were performed to calculate the similarity between purchase histories in consideration of the product classification system. The present invention has shown that dynamic time warping similarity is the most effective for measuring similarity between purchase histories among various methods for measuring similarity between sequences. Therefore, the similarity between two purchase histories is accurately and quickly measured based on dynamic time warping similarity in the following. suggest how to

Meanwhile, in the present invention, a segment tree is used to quickly calculate the similarity between the proposed purchase histories. A segment tree is a tree in which each node has information (eg, sum, minimum, maximum) about the sections represented by its child nodes. The segment tree is known to be effective in the problem of finding the minimum value within a specific range because information on each section is obtained and stored in advance.

3. Suggestion method

In this chapter, we propose a method for measuring the similarity between purchase histories considering the product classification system and an efficient calculation method for it.

3.1 Measurement of similarity between purchase histories considering the product classification system

크기의 2차원 배열 M 을 생성하고 모든 원소를 ∞로 초기화한다. 이후 M 의 각 원소 M [i][j]를 다음 수학식 2를 사용하여 업데이트한다. In the present invention, the dynamic time warping similarity, which is an existing representative method for measuring sequence similarity, is extended to reflect the product classification system. Suppose we are given two purchase histories s ₁ = < x ₁ , x ₂ , ..., x _n > and s ₂ = < y ₁ , y ₂ , ..., y _{m >.} Here, x _i and y _i represent the i- th purchased item in each purchase history. The proposed method is given when two purchase histories s ₁ and s ₂ are given.

Create a two-dimensional array M of size M and initialize all elements to ∞. Since each element M [i] [j] of M is then updated by using the equation (2).

[Equation 2]

M [ i ][ j ]= dist ( x _i , y _j )+min( M [ i - 1][ j ], M [ i ][ j - 1], M [ i - 1][ j - 1] )

In Equation 2 above, dist ( x _i , y _j ) represents a distance in consideration of the product classification system of the i- th item of s ₁ , x _i and the j- th item of s ₂ , y _{j .} Existing dynamic time warping degree of similarity is not considered a product classification system as _{dist (x i, y j)} = 1 if x _i = y _j a case _{dist (x i, y j)} = 0 , and x _i ≠ y _j applicable to the case In the above equation, the minimum of M [ i - 1][ j ], M [ i ][ j - 1], M [ i - 1][ j - 1] is added to dist ( x _i , y _j ) is s ₁ The cases in which comparison is continued by advancing only one item of s ₂ , the case of continuing comparison by advancing only one item of s ₂ , and the case of continuing comparison by advancing both items of s _{1 and s 2 one by one, respectively.} In the present invention, the value of dist ( x _i , y _j ) can have a value between 0 and 1 by reflecting the product classification system. dist ( x _i , y _j ) is defined as in Equation 3 below.

[Equation 3]

In the above Equation 3 T is a given product taxonomy _{tree, shortestPathLen (x i, y j} , T) is the length of the shortest path end-to-end node that represents a product x _i and y _j in T, longestPathLen (T) is T It represents the length of the path between the two terminal nodes that are the farthest in the . dist ( x _i , y _j ) has a value between 0 and 1, and has a smaller value as the two products are closer in the product classification tree, that is, the same subcategories are the same, and the further the two products are within the product classification tree, the smaller the value. , that is, it has a larger value as it belongs to different classifications. After obtaining all the values of M [ i ][ j ] through the above expression, the final value of M [ n ][ m ] becomes the distance between s ₁ and s _{2 .}

3.2 Efficient similarity calculation technique

In order to measure the degree of similarity between the previously proposed purchase history for dist (x _i, y _j) indicates the distance between the elements of the element y _j x _i and s ₂ s ₁ of the calculation should be repeated for all x _i and y _j. Therefore, the computational cost increases as the lengths of s ₁ and s _{2 increase.} Since longestPathLen ( T ) in dist ( x _i , y _j ) is a fixed value, it is very important to efficiently calculate shortestPathLen ( x _i , y _j , T ) for efficient similarity calculation. A simple way to find the shortestPathLen ( x _i , y _j , T ) for a given x _i and y _j is to find the path from the root node of T to x _i and y _j , respectively, and then to the node included in the two paths. It can be calculated very simply by comparing them one by one and finding the number of different nodes. However, this method has the disadvantage that the computation cost increases as the size of T increases. In particular, as the lengths of s ₁ and s ₂ become longer, this operation has to be repeated, so there is a problem in that the cost of calculating the overall similarity greatly increases.

Therefore, the present invention proposes a technique for very efficiently calculating the shortestPathLen ( x _i , y _j , T ) using a segment tree. The proposed method first traverses the product classification system tree once to create two one-dimensional arrays. The two arrays store the order of the nodes visited and the depth of each node, respectively. The newly proposed Equation 4 for calculating the shortestPathLen ( x _i , y _j , T ) using the segment tree generated through the array in which the visited node order is stored is as follows.

[Equation 4]

depth[LCA(x _i , y _j , T)] shortestPathLen ( x _i , y _j , T ) = depth [ x _i ]+ depth [ y _j ]-2

depth [ LCA ( x _i , y _j , T )]

In the above equation, LCA ( x _i , y _j , T ) is the closest common classification of the two products calculated by the segment tree, and depth[] is an array in which the depth of each node stored as a result of the translocation traversal is stored. The length of the path from x _i to LCA ( x _i , y _j , T ) is depth [ x _i ] -depth [ LCA ( x _i , y _j , T )], and from LCA ( x _i , y _j , T ) Since the length of the path up to _{y j} is depth [ y _j ] -depth [ LCA ( x _i , y _j , T )], the length of the shortest path of two x _i , y _{j becomes Equation 4 above.}

Then, when calculating the distance between two products in the product classification system tree, this segment tree is used to find the closest common classification of the two products and the depth information of the two products, and using this, shortestPathLen ( x _i , y _j , T ) calculate directly Therefore, when calculating shortestPathLen ( x _i , y _j , T ), the process of finding the path from the root node of T to x _i and y _j , and comparing the two paths is eliminated, so the total computation cost is greatly reduced.

4. Performance evaluation

In this chapter, the experimental results of the proposed similarity calculation method between purchase histories are presented. The proposed method was implemented using Python, and the experiment was performed on a PC in Windows 10 operating system environment equipped with Intel i7-5820 3.3GHz CPU and 8GB memory. The product classification system was used after collecting the actual classification system of Amazon, a representative online shopping mall, and building it in the form of a tree.

4.1 Evaluation of the effectiveness of the similarity measurement method

First, it was evaluated whether the proposed similarity measurement method considering the product classification system calculates the similarity between purchase histories more effectively than the existing sequence similarity measurement method. Table 4 shows the results of applying the proposed method to simple virtual purchase history data (similarity comparison experiment results).

Given three purchase histories s ₁ , s ₂ , and s ₃ have different products, s ₁ and s ₂ have similar product categories compared to s ₁ and s ₃ (eg, Coke and Sprite both belong to beverages) ) In this case, the existing dynamic time warping similarity calculates the distance between s ₁ and s _{2 and} the distance between s ₁ and s _{3 as} 6, but the proposed method calculates the distance between s ₁ and s _{2 and} the distance between s ₁ and s ₃ . is calculated as 1.5 and 5.375, respectively. Therefore, the proposed method calculates a more accurate distance compared to the existing method by giving a closer distance to products with similar product classifications, even when the two purchase histories are composed of different products. In addition, it was confirmed that there is no problem in application even if the purchase history lengths are different due to the characteristics of the dynamic time warping similarity.

[Table 4]

4.2 Performance evaluation of similarity calculation technique

Next, we evaluated how much the similarity calculation technique using the segment tree improves the calculation speed compared to the simple technique that does not use the segment tree. To this end, the degree of similarity was calculated 100 times, and the average time taken was measured. Randomly generated virtual purchase history data was used to calculate the similarity, and the experiment was performed by changing the length of the purchase history and the size of the product classification tree.

4 is a graph comparing the performance of a simple technique and a proposed technique to which a segment tree is applied according to a purchase history length. 5 is a graph comparing the performance of the simple technique and the proposed technique according to the number of products in the product classification system tree. Hereinafter, the performance of the technique proposed in the present invention will be described with reference to FIGS. 4 and 5 .

4 is a result of comparing the performance of the two methods while increasing the length of the purchase history while the product classification system tree is fixed. In both methods, the calculation time increases linearly as the length of the purchase history increases, but in the case of the proposed method, the shortest distance between two products in the product classification system tree can be quickly calculated using a pre-built segment tree, so the overall computational performance It can be seen that this has been greatly improved.

5 is a result of comparing the performance of the two methods while increasing the number of products in the product classification system tree from 7,000 to 25,000. In the case of the simple method, as the product classification system tree grows, the time to search for the shortest distance between two products increases, but it can be seen that the time does not increase because the proposed method uses a pre-built segment tree.

5. Conclusion

In the present invention, an effective method for measuring the similarity between two purchase histories in consideration of a product classification system and an efficient calculation technique thereof have been proposed. To this end, the dynamic time warping similarity, which is a representative method of measuring sequence similarity, was extended to reflect the distance between products within the product classification system.

In addition, we proposed a technique for quickly calculating the distance between products within the product classification system using a segment tree to efficiently calculate the proposed similarity measurement method. Experimental results showed that the proposed method measures the similarity between purchase histories more effectively than the existing method when a product classification system exists, and it is confirmed that it can be calculated very quickly using a segment tree.

6 is a flowchart illustrating a control method for determining a degree of similarity between sequences in consideration of an item classification system according to an embodiment of the present invention.

As shown in FIG. 6 , in the control method according to an embodiment of the present invention, a process of acquiring a first sequence (S210), a process of acquiring a second sequence (S230), and similarities of the first and second sequences It may include a process of determining whether or not (S250).

In steps S210 and S230, the first sequence may be composed of first items having an order, and the second sequence may be composed of second items having an order. For example, the first sequence may include item 1-1, item 1-2, and item 1-3 in order, and the second sequence may include item 2-1 and item 2-2 in order. , items 2-3 may be included. The number of items in the sequence may have a value of two or more.

For example, the first and second sequences may be product purchase history, the first purchase history may have the order of Sprite, Candy, and Jeans, and the second purchase history may have the order of Coke, Chocolate, and Skirt.

In step S250, it may be determined whether the first and second sequences are similar based on the first information and the second information.

The first and second items may correspond to a terminal node of the same item classification tree. The item classification tree is an item classification system having a tree structure, and branches from the root, which is the top node, to a lower hierarchy, and includes items (a to k) in a plurality of internal nodes (C1 to C8) and a terminal node that is the end of the plurality of internal nodes. ) (see FIG. 2(a)).

In order to determine whether the sequences are similar, one of <Item 1-1, Item 1-2, Item 1-3> of the first sequence and <Item 2-1 and Item 2-2 of the second sequence> It is possible to determine whether a sequence is similar by comparing one item among items and items 2-3>.

As a specific method, the first information may include information on the number of operations for converting one of the items of the first sequence into one of the items of the second sequence. In other words, the first information includes the number of operations for converting item 1-1 into item 2-1, item 2-2, item 2-3, and item 1-2 into item 2-1 and item 2-1. 2-2 items, the number of operations for converting items 2-3 into items, and the number of operations for converting items 1-3 into items 2-1, 2-2, and 2-3 can be included. have.

Also, the second information may include information about a length of a shortest path in the item classification tree between one of the first items and one of the second items.

Conventionally, information (0 or 1) on whether an item is identical to an item is used to determine the similarity between sequences, but this only sees the sameness between the items and does not include information on the relationship of nodes.

For example, according to the conventional method, correlation information between two items (Coke, Sprite) belonging to the same node and not identical and two items (Candy, Polos) of a different node are identically non-identical items. It has the information (0) meaning. However, it did not indicate that the items of (Coke, Sprite) were more related to each other than the items of (Candy, Polos) (see Fig. 2(a)).

The method proposed in the present invention uses the length of the shortest path between items in order to more precisely determine the similarity between sequences. It may be determined that the shorter the length of the shortest path between items, the higher the correlation information between items. For example, the path of items of (Candy, Polos) is 7, and the path of items of (Coke, Sprite) has a path of 2 (see FIG. 2(b) ). Therefore, it means that the items of (Coke, Sprite) are highly correlated compared to the items of (Candy, Polos).

Furthermore, the second information may further include information about a length of a longest path of items in the item classification tree. The method proposed in the present invention can measure the similarity between sequences by using the relative value of the correlation information between the items using the length of the main path of the items. For example, relative association information of items of (Coke, Sprite) has a value of 2/7 as the longest path length (7) of items in the item classification tree compared to the path length (2) of items of (Coke, Sprite). can

Also, in the method proposed by the present invention, depth information for each node of the item classification tree may be calculated and stored in advance in order to quickly calculate correlation information between items. Afterwards, when two items are selected, in order to calculate the length of the shortest path between the items, the shortest path between items based on the depth information, rather than calculating the path by comparing all nodes (root, internal node, terminal node) respectively. can calculate the length of Therefore, it is possible to prevent redundant operations and quickly provide the length of the shortest path between items.

If there is an item classification tree as shown in FIG. 2(a) , the length of a path from the root to each node may be defined as a node depth by assigning an order to each node. The node depth of the root is 0, the node depths of C1, C2, C3 are 1, the node depths of C4, C5, e, f, C6, C7 are 2, and a, b, c, d, g , h, I, and C8 may have a node depth of 3, and j and k may have a node depth of 4. The information of the node depth for each node may be calculated in the future and stored in the memory 1100 .

For example, if you want to find the length of the shortest path between items (Sandals, Jeans), determine the last shared lowest node (Fashion, C3) of Sandals(g) and Jeans(i) items, and Sandals(g) ), the length of the shortest path between items (Sandals, Jeans) can be determined through the sum of the path length between the item and the inner node of Fashion(C3) and the path length between the item and the inner node of Jeans(i) and Fashion(C3). . In this case, using the previously stored node depth information for each node, [depth(3) of Sandals(g) + depth(3) of Jeans(i) - depth(1) of 2*Fashion(C3) = 3 + 3) - 2*1 = 4], the length (4) of the shortest path between (Sandals, Jeans) items can be determined.

7 and 8 are block diagrams of an electronic device 1000 according to an exemplary embodiment.

The electronic device 1000 according to an embodiment includes a terminal, a device, an electronic device, a server, a smart phone, a tablet PC, a PC, a smart TV, a mobile phone, a personal digital assistant (PDA), a laptop, a media player, a micro server, a GPS It may be one of a global positioning system device, an e-book terminal, a digital broadcast terminal, a navigation device, a kiosk, an MP3 player, a digital camera, a home appliance, and other computing devices. Also, the electronic device may be a wearable device such as a watch, glasses, a hair band, and a ring having a display function and a data processing function. However, the present invention is not limited thereto, and the electronic device 1000 may include all kinds of devices capable of processing data and providing the processed data.

7 , the electronic device 1000 according to an embodiment may include a memory 1100 , an output unit 1200 , a communication unit 1500 , and a processor 1300 .

The processor 1300 performs each of the processes described above with reference to FIGS. 1 to 6 by the processor itself or controls the components of the electronic device 1000 in order to perform a control method for determining the degree of similarity between sequences in consideration of the item classification system. can be done by

Not all of the components shown in FIG. 7 are essential components of the electronic device 1000 , and the electronic device 1000 may be implemented with more components, and the electronic device 1000 with fewer components. may be implemented.

For example, as shown in FIG. 8 , the electronic device 1000 according to an embodiment includes a memory 1100 , an output unit 1200 , a processor 1300 , a sensing unit 1400 , and a communication unit 1500 . , an A/V input unit 1600 and a user input unit 1700 may be included.

*Memory (1100)

The memory 1100 may store a program for processing and controlling the processor 1300 , and may store information input to the electronic device 1000 or information output from the electronic device 1000 .

The memory 1100 may include a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (eg, SD or XD memory), and a RAM. (RAM, Random Access Memory) SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk , may include at least one type of storage medium among optical disks.

Programs stored in the memory 1100 may be classified into a plurality of modules according to their functions, for example, may be classified into a UI module 1110 , a touch screen module 1120 , a notification module 1130 , etc. .

The UI module 1110 may provide a specialized UI, GUI, or the like that interworks with the electronic device 1000 for each application.

The touch screen module 1120 may detect a touch gesture on the user's touch screen and transmit information about the touch gesture to the processor 1300 . The touch screen module 1120 according to an embodiment may recognize and analyze a touch code. The touch screen module 1120 may be configured as separate hardware including a controller.

The notification module 1130 may generate a signal for notifying the occurrence of an event in the electronic device 1000 . Examples of events generated in the electronic device 1000 include call signal reception, message reception, key signal input, schedule notification, and the like. The notification module 1130 may output a notification signal in the form of a video signal through the display unit 1210 , may output a notification signal in the form of an audio signal through the sound output unit 1220 , and the vibration motor 1230 . It is also possible to output a notification signal in the form of a vibration signal through For example, the notification module 1130 may generate a signal for outputting guide information based on the estimated lane information.

*Output unit (1200)

The output unit 1200 may output an audio signal, a video signal, or a vibration signal, and the output unit 1200 may include a display unit 1210 , a sound output unit 1220 , and a vibration motor 1230 . have.

The display unit 1210 displays and outputs information processed by the electronic device 1000 . Specifically, the display unit 1210 may output an image captured by the camera 1610 . The display 1210 may display a user interface for executing an operation related to a response in response to a user's input.

The sound output unit 1220 outputs audio data received from the communication unit 1500 or stored in the memory 1100 . Also, the sound output unit 1220 outputs a sound signal related to a function (eg, a call signal reception sound, a message reception sound, and a notification sound) performed by the electronic device 1000 .

*Processor (1300)

The processor 1300 generally controls the overall operation of the electronic device 1000 . For example, the processor 1300 executes programs stored in the memory 1100 , and thus the user input unit 1700 , the output unit 1200 , the sensing unit 1400 , the communication unit 1500 , and the A/V input unit 1700 . ) can be controlled in general.

*Sensing unit (1400)

The sensing unit 1400 may detect a state of the electronic device 1000 or a state around the electronic device 1000 , and transmit the sensed information to the processor 1300 .

The sensing unit 1400 includes a magnetic sensor 1410 , an acceleration sensor 1420 , a temperature/humidity sensor 1430 , an infrared sensor 1440 , a gyroscope sensor 1450 , and a position sensor. (eg, GPS) 1460 , a barometric pressure sensor 1470 , a proximity sensor 1480 , and at least one of an RGB sensor 1490 , but is not limited thereto. Since a function of each sensor can be intuitively inferred from the name of a person skilled in the art, a detailed description thereof will be omitted.

*Communication Department (1500)

The communication unit 1500 may include one or more components that allow the electronic device 1000 to communicate with another device (not shown) and a server (not shown). The other device (not shown) may be a computing device such as the electronic device 1000 or a sensing device, but is not limited thereto. For example, the communication unit 1500 may include a short-range communication unit 1510 , a mobile communication unit 1520 , and a broadcast receiving unit 1530 .

Short-range wireless communication unit 1510, Bluetooth communication unit, BLE (Bluetooth Low Energy) communication unit, short-range wireless communication unit (Near Field Communication unit), WLAN (Wi-Fi) communication unit, Zigbee (Zigbee) communication unit, infrared ( It may include an IrDA, infrared Data Association) communication unit, a Wi-Fi Direct (WFD) communication unit, an ultra wideband (UWB) communication unit, an Ant+ communication unit, and the like, but is not limited thereto.

The mobile communication unit 1520 transmits/receives a radio signal to and from at least one of a base station, an external terminal, and a server on a mobile communication network. Here, the wireless signal may include various types of data according to transmission/reception of a voice call signal, a video call signal, or a text/multimedia message.

The broadcast receiver 1530 receives a broadcast signal and/or broadcast-related information from the outside through a broadcast channel. The broadcast channel may include a satellite channel and a terrestrial channel. According to an embodiment, the electronic device 1000 may not include the broadcast receiver 1530 .

*A/V (Audio/Video) input unit (1600)*

The A/V (Audio/Video) input unit 1600 is for inputting an audio signal or a video signal, and may include a camera 1610 , a microphone 1620 , and the like.

The camera 1610 may obtain an image frame such as a still image or a moving image through an image sensor in a video call mode or a shooting mode. The image captured through the image sensor may be processed through the processor 1300 or a separate image processing unit (not shown).

The microphone 1620 receives an external sound signal and processes it as electrical voice data. For example, the microphone 1620 may receive an acoustic signal from an external device or a user. The microphone 1620 may receive a user's voice input. The microphone 1620 may use various noise removal algorithms for removing noise generated in the process of receiving an external sound signal.

*User input unit (1700)

The user input unit 1700 means a means for a user to input data for controlling the electronic device 1000 . For example, the user input unit 1700 includes a key pad, a dome switch, and a touch pad (contact capacitive method, pressure resistance film method, infrared sensing method, surface ultrasonic conduction method, integral type). There may be a tension measurement method, a piezo effect method, etc.), a jog wheel, a jog switch, and the like, but is not limited thereto.

Although the present invention has been described with reference to the embodiment shown in the drawings, this is merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Accordingly, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

Claims (8)

A control method for determining a degree of similarity between sequences in consideration of an item classification system, performed by an electronic device including a memory in which information on an item classification tree is previously stored and a processor controlling the memory, the control method comprising:
obtaining, by the electronic device, a first sequence including first items having an order;
obtaining, by the electronic device, a second sequence including second items having an order; and
determining, by the electronic device, whether the first and second sequences are similar to each other based on first information and second information by using the pre-stored information on the item classification tree;
The item classification tree is a top node, a plurality of internal nodes branched from the top node as a sub-class of the top node, and a sub-classification of each of the plurality of internal nodes and corresponds to the lowest node, each of the plurality of internal nodes Branched from and composed of a plurality of terminal nodes respectively corresponding to the first and second items,
The first information includes information about the number of operations for converting one of the first items into one of the second items,
The second information includes information about a length of a shortest path in the itemized tree between one of the first items and one of the second items, wherein the length of the shortest path in the itemized tree is two It is the sum of the paths between the internal node corresponding to the closest common classification in the terminal node and the two terminal nodes,
In the process of determining whether the first and second sequences are similar, the shorter the length of the shortest path in the item classification tree between one of the first items and one of the second items, the shorter the length of the first items. determining a high degree of similarity between one of the first items and one of the second items, and determining a higher degree of similarity between the first and second sequences as the similarity between one of the first items and one of the second items increases. characterized by,
A control method for determining the similarity between sequences considering the item classification system. delete According to claim 1,
wherein the second information further includes information about a length of a longest path of items in the item classification tree;
A control method for determining the similarity between sequences considering the item classification system. According to claim 1,
The length of the shortest path in the item classification tree between one of the first items and one of the second items is determined based on pre-stored depth information for each node of the item classification tree,
A control method for determining the similarity between sequences considering the item classification system. a memory for storing information about the item classification tree; and
A processor for controlling the memory,
The processor obtains a first sequence composed of first items having an order, obtains a second sequence composed of second items having an order, and uses the pre-stored information about the item classification tree to obtain a first Control to determine whether the first and second sequences are similar based on the information and the second information,
The item classification tree is a top node, a plurality of internal nodes branched from the top node as a sub-class of the top node, and a sub-classification of each of the plurality of internal nodes and corresponds to the lowest node, each of the plurality of internal nodes Branched from and composed of a plurality of terminal nodes respectively corresponding to the first and second items,
The first information includes information about the number of operations for converting one of the first items into one of the second items,
the second information includes information about a length of a shortest path in the itemized tree between one of the first items and one of the second items, wherein the length of the shortest path in the itemized tree is two It is the sum of the paths between the internal node corresponding to the closest common classification in the terminal node and the two terminal nodes,
In the process of determining whether the first and second sequences are similar, the shorter the length of the shortest path in the item classification tree between one of the first items and one of the second items, the shorter the length of the first items. determining a high degree of similarity between one of the first items and one of the second items, and determining a higher degree of similarity between the first and second sequences as the similarity between one of the first items and one of the second items increases. An electronic device for determining a degree of similarity between sequences in consideration of an item classification system. delete 6. The method of claim 5,
wherein the second information further includes information about a length of a longest path of items in the item classification tree;
An electronic device for determining the similarity between sequences in consideration of the item classification system. 6. The method of claim 5,
and a length of a shortest path in the item classification tree between one of the first items and one of the second items is determined based on depth information about each node of the item classification tree pre-stored in the memory. to do,
An electronic device for determining the similarity between sequences in consideration of the item classification system.

Images