CN102033886B - Fabric retrieval method and system - Google Patents

Abstract

The invention relates to a fabric retrieval method, which comprises the following steps: a server receives a user query request from a client; the server side calculates the final weight value of each attribute of the fabric to be retrieved in the query request; the server side groups the historical records and calculates a characteristic record representing each group; then, the server side determines the group to which the feature record with the highest matching value with the query request belongs; the server side further determines the history record with the highest matching value with the query request and sends the history record to the client side. The invention also discloses a system using the method. By utilizing the method and the system disclosed by the invention, a garment designer and a purchaser can select a proper fabric for a new product in a most efficient mode, so that the effective management is provided for the sample data of the fabric, the research and development process of the new product is accelerated, and the production capacity and the competitive capacity of the textile industry and the garment industry are further enhanced.

Description

A fabric retrieval method and system

technical field

The invention relates to the field of data retrieval, more specifically, to a fabric retrieval method and system.

Background technique

In a rapidly changing society, the development of new products in the apparel industry requires a shorter life cycle, faster product response and release. At the same time, more and more uncertain customer needs and purchasing impulses have brought enormous pressure to apparel companies. Opportunities and Challenges. Fashion designers and buyers have to constantly develop new products to meet the needs of the market. In order to meet these challenges, clothing companies have to accelerate their new product development process NPD (new product development). NPD in the apparel industry includes market research, trend analysis, sourcing, design and final garment shaping.

In the NPD process, there are two links: the designer designs a new product, and the designer forms the fabric specification of the new product according to the appearance and functional requirements of the new product, such as color, style, whether it is waterproof, washing method, etc. These new The fabric specification of the product is also the collection of the attribute values of the fabric required by the new product; then the buyer selects the appropriate fabric according to the fabric specification required by the new product designed by the designer. In the process of selecting a suitable fabric, the buyer should not only consider the fabric specifications provided by the designer, but also consider the relevant factors of the supply chain that produces the product: such as suppliers, costs, inventory, etc. Due to the different experiences of designers and buyers, different views on fabrics, different focus of attention, different scales of grasping attributes, or insufficient communication between designers and buyers, the fabrics selected by buyers are often not available. If the designer agrees, then the buyer and the designer need to communicate, and then the buyer has to re-select the fabric. And because there are tens of thousands of samples of fabrics, the process of selecting fabrics often needs to be repeated many times until the fabrics that both designers and purchasers agree with are selected.

The efficiency of the above two links is very low, which greatly affects the efficiency of NPD and the correctness of fabric selection. Therefore, without the support of successful similar experience in the past, designers and purchasers can only simply design the products required by customers and select the required fabrics according to their instinct, experience and awareness. That is to say, the NPD process lacks the support of knowledge.

At the same time, the NPD team needs to use information management systems such as ERP (Enterprise Resource Planning) to manage business data and information. In the process of repeatedly evaluating and selecting fabrics, a large amount of useful data will be generated. However, due to the lack of a systematic method to store and manipulating all relevant data, resulting in very inefficient data management for the NPD process.

Therefore, there is a need for a fabric retrieval method and system to overcome the above-mentioned defects in the prior art.

Contents of the invention

The technical problem to be solved by the present invention is to propose a fabric retrieval method and system for the inefficient fabric data management and the lack of knowledge support in the selection process existing in fabric selection.

One of the technical schemes used by the present invention to solve the technical problems is to provide a fabric retrieval method, comprising the steps of:

The server side receives a user query request from the client, wherein the user query request includes historical record grouping information, the value of at least one attribute of the fabric to be retrieved, and the weight level and label corresponding to each attribute in the at least one attribute; The historical record grouping information includes the value that the number of historical records in each group of historical records must meet and the value that the historical record similarity coefficient in each group of historical records must meet;

The server end calculates the final weight value of each attribute of the fabric to be retrieved in the query request according to the weight level and the label of each attribute in the received query request;

According to the value that the number of historical records in each group of historical records in the historical record grouping information must satisfy and the value that the historical record similarity coefficient in each group of historical records must satisfy, the server side performs the historical records stored in the historical record database Group, and calculate the characteristic records representing each group;

The server side calculates each The matching value of the feature record and the query request, determining the group to which the feature record with the highest matching value with the query request belongs;

According to the value of each attribute in the query request and calculated by the server, the server end calculates the final weight value of each attribute and each history in the group to which the feature record with the highest matching value with the query request belongs. The value of each attribute of the record, calculate the matching value of each historical record in the group to which the feature record with the highest matching value of the query request and the query request is determined, determine the historical record with the highest matching value of the query request, and put the The history is sent to the client.

In the fabric retrieval method of the present invention, the label of each attribute is a non-repetitive continuous natural number starting from 1, and the final weight value of each attribute of the fabric to be retrieved in the calculation query request includes the following step:

S21) Calculate the initial weight value w _i of each attribute in the query request, according to $w_i=\frac{1}{\mathrm{no}}\underset{j=i}{\overset{\mathrm{no}}{\mathrm{\Σ}}}\frac{1}{j},i=\mathrm{1,2}...\mathrm{no}$ To calculate the initial weight value of each attribute, where i represents the label of the attribute, and n represents the total number of attributes of the fabric to be retrieved in the query request;

S22) Add the initial weight values of the attributes with the same weight level and take the average value to obtain the final weight w(A _i ) of the attribute with the weight level.

In the fabric retrieval method of the present invention, the similarity coefficient of the historical records is: the ratio of the number of common attributes that any two historical records have to the number of all attributes of each historical record in the two historical records and the average value of .

In the fabric retrieval method of the present invention, the grouping of the historical records satisfies the following conditions:

(1) If C _i , C _j ∈ q _k , then max(θ(C _i , C _j ))≥σ, indicating the minimum value that must be satisfied by the maximum similarity coefficient in each group of historical records;

$<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; q</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; q</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; \&Greater\; Equal;</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}<span\; class="notranslate">\; ;</span>$

(3) If C _i ∈ q _k then $C_i\&NotElement;q_{\mathrm{the\; s}},$ k≠s, q _k , q _s ∈ Q;

in, $\mathrm{\&theta;}=(C_i\&cap;C_j)=\frac{1}{2}(\frac{\left|A\left(C_i\right)\right|\&cap;A\left(C_j\right)|}{\left|A\left(C_i\right)\right|}+\frac{|A\left(C_i\right)\&cap;A\left(C_j\right)|}{\left|A\left(C_j\right)\right|})$ is the similarity coefficient of historical records C _i and C _j , σ is the value that the similarity coefficient of historical records in each group of historical records in the historical record grouping information must satisfy, and β is the history of each group of historical records in the historical record grouping information The number of records must meet the value, C=(C ₁ , C ₂ ,...,C _j ,...C _m ) means the collection of historical records, A=(A ₁ , A ₂ ,...,A _j ,...A _n ) represents the set of all attributes, A(C _i )=(A ₁ , A ₂ ,...,A _j ,...A _k ) represents the set of all attributes of the historical record C _i , Q=(q ₁ , q ₂ , . . . , q _m ) represents the set of historical record groups, and |A(C _i )| represents the number of all attributes in the historical record C _i .

In the fabric retrieval method according to the present invention, the feature record of each group is a record composed of the average value of all historical records in the group on the same attribute, and calculating the feature record of each group includes the following steps:

S31) q _j (A _i ) represents the average value of all historical records in group q _j on attribute A _i , using $\stackrel{\&OverBar;}{q_j}\left(A_i\right)=\frac{1}{\left|q_j^i\right|}\underset{\&ForAll;C_k\&Element;q_j^i}{\mathrm{\&Sigma;}}{v}_k\left(A_i\right)$ Calculate the average value of all historical records in group q _j on each attribute, v _k (A _i ) represents the value of attribute A _i of historical record k, q _j ⁱ represents the set of all historical records in group q _j , |q _j ⁱ |Indicates the number of all historical records in the group q _j ;

为分组q_j的特征记录， $v_{C_{q_j}^r}\left(A_i\right)=\stackrel{\&OverBar;}{q_j}\left(A_i\right).$ S32)

is the feature record of group q _j , $v_{C_{q_j}^r}\left(A_i\right)=\stackrel{\&OverBar;}{q_j}\left(A_i\right).$

In the fabric retrieval method of the present invention, the matching value of each feature record and the query request is calculated as: $(C_{\mathrm{input}},C_{q_j}^r)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_j}^r}\left(A_i\right))),$ in $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_j}^r}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{C_{q_j}^r}\left(A_i\right)|}^2,$ C _input represents the query request, Represents the feature record of group q _j .

In the fabric retrieval method of the present invention, the matching value of each historical record in the group to which the feature record with the highest matching value is calculated is: $(C_{\mathrm{input}},C_j)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_j}\left(A_i\right))),$ in $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{c_j}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{c_j}\left(A_i\right)|}^2,$ C _j represents the history record in the group to which the feature record with the highest matching value belongs, and C _input represents the query request.

The second technical solution used by the present invention to solve the technical problems is to provide a fabric retrieval system, including a client and a server, and the client includes:

Fabric database: used to store fabric-related business data;

Multidimensional data management module: used to receive the query request input by the user and send the request to the server, and use the star schema to manage the data in the fabric database and provide fabric business-related operations;

The server side includes:

Retrieval module: used to calculate the final weight value of each attribute of the fabric to be retrieved in the query request sent by the multidimensional data management module, group the historical records in the historical record database and calculate the feature records of each group, calculate each A characteristic record and the matching value of the query request, determine the grouping where the characteristic record with the highest matching value is located in the query request, calculate the matching value of each historical record and the query request in the grouping where the characteristic record with the highest matching value is located, and determine the matching value with the query request Querying the historical record with the highest matching value, and sending the historical record to the multidimensional data management module of the client;

History database: used to store history.

In the fabric retrieval system of the present invention, the retrieval module includes:

Historical record grouping unit: used to group historical records in the historical record database, and calculate characteristic records representing each group;

Attribute weight calculation unit: used to calculate the final weight value of each attribute of the fabric to be retrieved in the query request;

Group search unit: used for calculating the final weight value of each attribute of the fabric to be retrieved in the query request calculated by the attribute weight calculation unit, the value of each attribute in the query request and the historical record grouping The unit calculates the value of each attribute of the feature record, calculates the matching value of the query request and each feature record, and determines the group to which the feature record with the highest matching value with the query request belongs;

History record lookup unit: used for calculating the final weight value of each attribute of the fabric to be retrieved in the query request calculated by the attribute weight calculation unit, the value of each attribute in the query request and the group search unit to determine The value of the attribute of each historical record in the group to which the feature record with the highest matching value belongs, and calculate the query request and each historical record in the group to which the feature record with the highest matching value determined by the group search unit belongs The matching value of is determined to determine the historical record with the highest matching value of the query request.

Implementing the technical solution of the present invention has the following beneficial effects: in the fabric selection process of new products, support is provided for the fabric selection of new products by referring to existing historical records, the efficiency of fabric selection for new products is provided, and at the same time, it provides support for the fabric selection of fabrics. Sample data provides effective management and speeds up the process of new product development.

Description of drawings

Fig. 1 is a flow chart of a fabric retrieval method provided by a preferred embodiment of the present invention;

Fig. 2 is a schematic structural diagram of a fabric retrieval system provided by a preferred embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

In the embodiment of the present invention, the notation of symbol is as follows:

C=(C ₁ , C ₂ , . . . , C _j , . . . C _m ) represents the collection of all historical records in the knowledge base;

A=(A ₁ , A ₂ , . . . , A _j , . . . A _n ) represents the set of all attributes;

A(C _i )=(A ₁ , A ₂ ,..., A _j ,...A _k ) represents the set of all attributes of the historical record C _i ;

|M| indicates the number of elements in the set M;

v _i (A _j ) represents the value of attribute A _j of history record i;

w(A _i ) represents the final weight value of attribute A _i ;

$\mathrm{<span\; class="notranslate">\; sim</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ;</span>$

Q=(q ₁ , q ₂ , . . . , q _m ) represents a collection of historical record groups.

Fig. 1 is a flowchart of a fabric retrieval method provided by a preferred embodiment of the present invention. As shown in Fig. 1, in step S1, the server side receives a user query request from the client side.

The user query request includes historical record grouping information, the value of at least one attribute of the fabric to be retrieved, and the weight level and label corresponding to each attribute in the at least one attribute; the historical record grouping information includes the number of historical records in each group of historical records. Satisfied values and values that must be satisfied by the historical record similarity coefficient in each set of historical records.

In the specific implementation process, firstly, the designer forms the fabric specifications of the new product according to the appearance and functional requirements of the new product, such as color, style, whether it is waterproof, washing method, etc. These fabric specifications are also a collection of fabric attribute values, and then , the buyer extracts the fabric specifications of the new product designed by the designer to select the appropriate fabric for the new product. In the process of selecting a suitable fabric, the buyer should not only consider the fabric specifications provided by the designer, but also consider the relevant factors of the supply chain that produces the product: such as suppliers, costs, etc. Buyers set a weight level and label for the fabric attribute specifications designed by designers and other related attribute specifications in the supply chain based on their experience, and determine the value that the number of historical records in each group of historical records must meet The value that must be satisfied by the historical record similarity coefficient in each group of historical records, and all these information are input to the client as a query request.

Subsequently, in the next step S2, the server calculates the final weight value of each attribute of the fabric to be retrieved in the query request according to the weight level and the label of each attribute in the received query request.

In an embodiment of the present invention, the label of each attribute is a non-repetitive continuous natural number starting from 1, and the calculation of the final weight value of each attribute of the fabric to be retrieved in the query request includes the following steps:

S21) Calculate the initial weight value w _i of each attribute in the query request, according to $w_i=\frac{1}{\mathrm{no}}\underset{j=i}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}\frac{1}{j},i=\mathrm{1,2},...\mathrm{no}$ To calculate the initial weight value of each attribute, where i represents the label of the attribute, and n represents the total number of attributes of the fabric to be retrieved in the query request;

S22) Add the initial weight values of the attributes with the same weight level and take the average value to obtain the final weight w(A _i ) of the attribute with the weight level.

For example, the query request C _input has 11 attributes in total, each attribute has a label from 1 to 11 and five weight levels: very important, important, not very important, relevant and irrelevant. As shown in Table 1:

Attributes level label Composition of the fabric Very important 1 Type of product Very important 2 fabric softness Very important 3 fabric structure important 4

How to maintain the fabric important 5 season of use important 6 color/pattern not so important 7 supplier address not so important 8 Fabric width/weight relevant 9 Dyeing method relevant 10 cleaning method irrelevant 11

Table 1

Then: the weight of the attribute labeled 1 is: ${\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="H2009101792825E00012.png"\; state="\{\{state\}\}">w</figure-callout>}}_1=\frac{1}{11}(1+\frac{1}{2}+\frac{1}{3}+\frac{1}{4}+\frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}+\frac{1}{9}+\frac{1}{10}+\frac{1}{11})=0.275;$

The weight of the attribute labeled 2 is: ${\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="H2009101792825E00012.png"\; state="\{\{state\}\}">w</figure-callout>}}_2=\frac{1}{11}(\frac{1}{2}+\frac{1}{3}+\frac{1}{4}+\frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}+\frac{1}{9}+\frac{1}{10}+\frac{1}{11})=0.184;$

The weight of the attribute labeled 3 is: $w_3=\frac{1}{11}(\frac{1}{3}+\frac{1}{4}+\frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}+\frac{1}{9}+\frac{1}{10}+\frac{1}{11})=0.138;$

By analogy, the weights of attributes ranked 4th to 11th are: 0.108, 0.085, 0.067, 0.052, 0.039, 0.027, 0.017, and 0.008.

Finally, the initial weights of attributes with the same rank are summed and averaged to get the final weights:

The three attributes of fabric composition, fabric type, and fabric softness have the same level, then the final weights of these three attributes are:

$\frac{1}{3}(0.275+0.184+0.138)=0.199.$

Similarly, the final weights of the three attributes of fabric structure, fabric maintenance, and season of use are:

$\frac{1}{3}(0.108+0.085+0.067)=0.087.$

The final weight of the two attributes Color/Pattern, Supplier Address is: 0.045.

The final weight of the two attributes Fabric Width/Weight, Dyeing Method: 0.024.

The final weight of this attribute of washing method is: 0.008.

In the embodiment of the present invention, as long as the label of each attribute of the historical record is a non-repeating continuous natural number starting from 1, it is not important which attribute corresponds to which number.

Subsequently, in the next step S3, the server end stores in the historical records according to the value that the number of historical records in each group of historical records in the historical record grouping information must satisfy and the value that the historical record similarity coefficient in each group of historical records must satisfy. The historical records in the record database are grouped, and the characteristic records representing each group are calculated.

In the embodiment of the present invention, the similarity coefficient of the historical records is: the average value of the sum of the ratio of the number of common attributes of any two historical records to the number of all attributes of each of the two historical records.

For example: historical record C _i has 16 attributes, historical record C _j has 12 attributes, and these two historical records have 5 common attributes, then the similarity coefficient of these two historical records $\mathrm{\&theta;}=(C_i\&cap;C_j)=\frac{1}{2}(\frac{|A\left(C_i\right)\&cap;A\left(C_j\right)|}{\left|A\left(C_i\right)\right|}+\frac{|A\left(C_i\right)\&cap;A\left(C_j\right)|}{\left|A\left(C_j\right)\right|})=\frac{1}{2}(\frac{5}{16}+\frac{5}{12}).$

The grouping of historical records meets the following conditions:

(1) If C _i , C _j ∈ q _k , then max(θ(C _i , C _j ))≥σ, σ is the value that the similarity coefficient of historical records in each group of historical records in the historical record grouping information must satisfy. Here σ is the minimum value that must be satisfied by the maximum similarity coefficient in each group;

(2) and ${\&ForAll;q}_j,\left|q_j\right|\&Greater\; Equal;\mathrm{\&beta;},$ β is the value that must be satisfied by the number of historical records in each group of historical records in the historical record grouping information, where β is the minimum number of historical records that must be satisfied by each group;

(3) If C _i ∈ q _k then $C_i\&NotElement;q_{\mathrm{the\; s}},$ k≠s, q _k , q _s ∈ Q, means that a historical record can only belong to one group.

After completing the grouping of existing historical records according to the above (1), (2), and (3), step S3 further calculates the feature records representing each group. The feature records of each group are all historical records in the group in the same attribute The records composed of the average value above, the calculation of the feature records of each group includes the following steps:

S31) q _j (A _i ) represents the average value of all historical records in group q _j on attribute A _i , using $\stackrel{\&OverBar;}{q_j}\left(A_i\right)=\frac{1}{\left|q_j^i\right|}\underset{\&ForAll;C_k\&Element;q_j^i}{\mathrm{\&Sigma;}}{v}_k\left(A_i\right)$ Calculate the average value of all historical records in group q _j on each attribute, v _k (A _i ) represents the value of the i-th attribute of historical record k, q _j ⁱ represents the set of all historical records in group q _j , |q _j ⁱ |Indicates the number of all historical records in the group q _j ;

S32) C _j ^r is the feature record of group q _j , $v_{C_j^r}\left(A_i\right)=\stackrel{\&OverBar;}{q_j}\left(A_i\right).$

Subsequently, in the next step S4, the server side calculates each A matching value of the feature record and the query request, determining the group to which the feature record with the highest matching value with the query request belongs.

表示分组q_j的特征记录。In step S4, the matching value of each feature record and query request is calculated as: $(C_{\mathrm{input}},C_{q_j}^r)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_j}^r}\left(A_i\right))),$ in $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_j}^r}\left(A_i\right))={\left|v_{\mathrm{input}}(A_i-v_{C_{q_j}^r}\left(A_i\right))\right|}^2,$ C _input represents the query request,

Represents the feature record of group q _j .

C_input与分组q₃的特征记录的匹配值(C_input，C_q3 ^r)、C_input与分组q₄的特征记录的匹配值

C_input与分组q₅的特征记录的匹配值

Assume that five groups are generated after grouping the historical records in step S3, namely Q=(q ₁ , q ₂ , q ₃ , q ₄ , q ₅ ). Then the matching value of the query request C _input and the feature record of group q ₁ is: $\left(C_{\mathrm{input}}{,C}_{q_1}^r\right)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_1}^r}\left(A_i\right))),$ in $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_1}^r}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{C_{q_1}^r}\left(A_i\right)|}^2,$ w(A _i ) is the weight corresponding to attribute A _i . In the same way, calculate the matching value of the query request C _input and the feature record of the group q ₂

The matching value of C _input and the feature record of group q ₃ (C _input , C _q3 ^r ), the matching value of C _input and the feature record of group q ₄

The matching value of C _input and the feature record of group q ₅

After completing the calculation of the matching value of the query request and the feature records of each group of historical records, determine the group to which the feature record with the highest matching value of the query request belongs, assuming it is group q ₃ .

Subsequently, in the next step S5, the server calculates the final weight value of each attribute and each attribute in the group to which the feature record with the highest matching value of the query request belongs according to the value of each attribute in the query request. For the value of each attribute of the historical record, calculate the matching value of each historical record in the group to which the feature record with the highest matching value matches the query request and the query request, determine the historical record with the highest matching value with the query request, and put the historical record Records are sent to the client.

In step S5, the matching value of the query request and each historical record in the group to which the feature record with the highest matching value belongs is calculated: $(C_{\mathrm{input}},C_j)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_j}\left(A_i\right))),$ in $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{c_j}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{c_j}\left(A_i\right)|}^2,$ C _j represents the history record in the group to which the feature record with the highest matching value belongs, and C _input represents the query request.

Because the group with the highest matching value selected in step S4 is q ₃ , step S5 will determine a historical record with the highest matching value in group q ₃ as the result of this query

Assume there are 6 history records C ₁ , C ₂ , C ₃ , C ₄ , C ₅ and C ₆ in the group q ₃ with the highest matching value. Then the matching value of the query request C _input and the historical record C ₁ is: $(C_{\mathrm{input}},C_1)=\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}w\left(A_i\right)*(1-\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_1}\left(A_i\right))),$ Where w(A _i ) is the weight corresponding to attribute A _i , $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_1}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{C_1}\left(A_i\right)|}^2.$ Similarly, calculate the matching value (C _input , C ₂ ) of the query request C _input and the historical record C ₂ , the matching value (C _input , C ₃ ) of the historical record C ₃ , and the matching value (C input ) of the historical record C ₄ (C _input , C ₄ ), the matching value (C _input , C ₅ ) with the historical record C ₅ , and the matching value (C _input , C ₆ ) with the historical record C ₆ . Take the historical record with the highest matching value, assuming it is C ₅ , and send it to the client as the result of this query.

In a specific implementation process, after step S5, the purchaser provides the fabrics corresponding to the selected historical records to the designer, and the designer can know the properties of these fabrics according to the previous historical records. The designer then analyzes and evaluates the fabric provided by the buyer. The evaluation may have the following situations:

(1) The fabric provided by the buyer meets the designer's requirements. In this case, the designer's opinion and the buyer's opinion have reached an agreement, and the fabric provided by the buyer is the fabric required by the new product designed by the designer.

(2) The fabric provided by the buyer does not meet the needs of the designer. The designer and the buyer discuss together and analyze the previous historical records. It is found that for the new design of the designer, the retrieved fabric is better than the fabric required by the designer. More superior, then, the designer changes his design requirements to meet the specifications of the fabric. The fabric provided by the buyer is also the fabric required by the final designer for the new product.

(3) The fabric provided by the buyer does not meet the needs of the designer. The designer and the buyer discuss together. The result of the discussion is that the designer may need to modify the design requirements, or the buyer may need to modify some other specifications other than the design requirements. , or both designers and buyers need to revise some of their specifications for this new product. Then, the buyer re-uses the retrieval method described in the present invention to search for the fabric of the new product according to the new specification requirements after the new product modification.

In practice, due to the different experiences of designers and buyers, different views on fabrics, different focus of attention, different scales of grasping attributes, or insufficient communication between designers and buyers, the retrieval method described in the present invention The process may go through several iterations before finding a fabric that both the final designer and buyer are happy with.

FIG. 2 is a schematic structural diagram of a fabric retrieval system provided by a preferred embodiment of the present invention. As shown in FIG. 2 , it includes: a client terminal 1 and a server terminal 2 . The client 1 includes a multi-bit data management module 11 and a fabric database 12 . The server end 2 includes a retrieval module 21 and a historical record database 22 . The retrieval module 21 includes: an attribute weight calculation unit 211 , a historical record grouping unit 212 , a group search unit 213 and a historical record search unit 214 .

In an embodiment of the present invention, the fabric database 12 is used to store fabric-related business data, and the stored data not only includes static attribute data related to the fabric itself, but also includes supply chain-related data, such as: supplier information, produced product information, etc. Information, historical transaction records, customer information, costs, inventory and employee related information, etc., all data related to clothing business. The multidimensional data management module 12 then organizes and manages the data in the fabric database into a star schema (a star schema is a collection of some tables in a relational database designed according to the dimension modeling principle), and this model can speed up when different data How responsive the source is when fetching data. A star schema creates a multidimensional space outside of a relational table, allowing single-dimensional tables to be connected to each other. At the same time, the multidimensional data management module 12 provides operations related to the fabric business: including operations such as query, insertion, deletion, and update, and uses these operations to realize data management. These operations are implemented through Structured Query Language (SQL). Users use the corresponding functions on the multi-dimensional data management module 12 to complete their own work according to the business they are responsible for.

For example, in the embodiment of the present invention, firstly, the designer is in charge of designing new products. Then, the designer forms the fabric specifications of the new product according to the appearance and functional requirements of the new product, such as color, style, whether it is waterproof, washing method, etc. These fabric specifications are also a collection of fabric attribute values. The value of is input in the multidimensional data management module 12, then the multidimensional data management module 12 saves these fabric specifications in the fabric database 11 with relational table format. Next, the business of the buyer is to find out the appropriate fabric for the new product according to the demand specification of the new product designed by the designer. Then, the buyer extracts the fabric specifications of the new product designed by the designer in the multidimensional data management module 12 to select a suitable fabric for the new product. In the process of selecting a suitable fabric, the buyer should not only consider the fabric specifications provided by the designer, but also consider the relevant factors of the supply chain that produces the product: such as suppliers, costs, inventory, etc. Buyers set a weight level and label for the fabric attribute specifications designed by designers and other related attribute specifications in the supply chain based on their experience, and determine the value that the number of historical records in each group of historical records must meet The value that must be satisfied with the historical record similarity coefficient in each group of historical records, all these information are entered into the multidimensional data management module 12 as a query request to select the fabric, and the multidimensional data management module 21 triggers the retrieval of the server end 2 when receiving the fabric selection command Module 21 to select the appropriate fabric.

The retrieval module 21 of the server end 2 receives the query request from the multidimensional data management module 21, first, the attribute weight calculation unit 211 calculates the weight of each attribute of the query request according to the attribute weight calculation method in the retrieval method of the present invention; then, The historical record grouping unit 212 groups the historical records in the historical record database 22 according to the grouping method in the retrieval method of the present invention, and calculates the representative feature records according to the method for calculating representative feature records for the grouped historical records in the retrieval method of the present invention. The feature record of each group; then the group search unit 213 determines the highest matching value with the query request according to the method of calculating the matching value of the query request and the feature record of each group in the historical record database in the retrieval method of the present invention. The grouping to which the characteristic record belongs; finally, the historical record search unit 214 utilizes the method of calculating the query request in the retrieval method of the present invention and the matching value of each historical record in the grouping to which the characteristic record with the highest matching value belongs to determine and query Request the history record with the highest matching value and send this history record to client 1.

In the specific implementation process, if the final designer and the buyer reach an agreement on the retrieved fabric, that is, the retrieved fabric is the final fabric of the new product designed by the designer, then this retrieval will be stored as a historical record in the historical record database 22 middle.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.

Claims (8)

1. A method of retrieving a fabric, comprising the steps of:

the method comprises the steps that a server receives a user query request from a client, wherein the user query request comprises historical record grouping information, a value of at least one attribute of a fabric to be retrieved and a weight grade and a label corresponding to each attribute in the at least one attribute; the historical record grouping information comprises values which are required to be met by the number of the historical records in each group of historical records and values which are required to be met by the historical record similarity coefficient in each group of historical records;

the server side calculates a final weight value of each attribute of the fabric to be retrieved in the query request according to the weight grade and the label of each attribute in the received query request;

the server side groups the historical records stored in the historical record database according to the value which is required to be met by the number of the historical records in each group of the historical records in the historical record grouping information and the value which is required to be met by the similarity coefficient of the historical records in each group of the historical records, and calculates the characteristic record representing each group;

the server side calculates a matching value of each feature record and the query request according to the value of each attribute in the query request, the final weight value of each attribute calculated by the server side and the value of each attribute of each feature record, and determines a group to which the feature record with the highest matching value with the query request belongs;

the server side calculates the matching value of each history record in the group to which the feature record with the highest matching value with the query request belongs and the query request according to the value of each attribute in the query request, the final weight value of each attribute calculated by the server side and the value of each attribute of each history record in the group to which the feature record with the highest matching value with the query request belongs, determines the history record with the highest matching value with the query request, and sends the history record to the client side;

wherein the label of each attribute is a non-repeating continuous natural number starting from 1, and the calculating the final weight value of each attribute of the fabric to be retrieved in the query request includes the following steps:

s21) calculating an initial weight value w of each attribute in the query request_iAccording to

Calculating an initial weight value of each attribute, wherein i represents the label number of the attribute, and n represents the total number of the attributes of the fabric to be searched for according to the query request;

s22) adding the initial weight values of the attributes having the same weight levelThen, the final weight w (A) of the attribute with the weight grade is obtained by taking the average value_i)。

2. The method of claim 1, wherein the similarity coefficient of the history record is: the average value of the sum of the ratios of the number of common attributes of any two historical records to the number of all the attributes of each historical record in the two historical records.

3. The method of claim 2, wherein the grouping of the history records satisfies the following condition:

(1) if C is present_i，C_j∈q_kThen max (θ (C)_i，C_j) σ) is larger than or equal to σ, and represents the minimum value which the maximum similarity coefficient in each group of history records must satisfy;

(2) <math> <mrow> <mo>&ForAll;</mo> <msub> <mi>q</mi> <mi>j</mi> </msub> <mo>,</mo> <mo>|</mo> <msub> <mi>q</mi> <mi>j</mi> </msub> <mo>|</mo> <mo>&GreaterEqual;</mo> <mi>&beta;</mi> <mo>;</mo> </mrow> </math>

(3) if C is present_i∈q_kThen

k≠s，q_k，q_s∈Q；

Wherein, <math> <mrow> <mi>&theta;</mi> <mo>=</mo> <mfenced open='(' close=')'> <mtable> <mtr> <mtd> <msub> <mi>C</mi> <mi>i</mi> </msub> <mi>I</mi> </mtd> <mtd> <msub> <mi>C</mi> <mi>j</mi> </msub> </mtd> </mtr> </mtable> </mfenced> <mo>=</mo> <mfrac> <mn>1</mn> <mn>2</mn> </mfrac> <mrow> <mo>(</mo> <mfrac> <mfenced open='|' close='|'> <mtable> <mtr> <mtd> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mi>I</mi> </mtd> <mtd> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mtd> </mtr> </mtable> </mfenced> <mrow> <mo>|</mo> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> <mo>+</mo> <mfrac> <mfenced open='|' close='|'> <mtable> <mtr> <mtd> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mi>I</mi> </mtd> <mtd> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mtd> </mtr> </mtable> </mfenced> <mrow> <mo>|</mo> <mi>A</mi> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> <mo>)</mo> </mrow> </mrow> </math> for history C_iAnd C_jσ is a value that the history similarity coefficient in each group of the history grouping information should satisfy, β is a value that the number of histories in each group of the history grouping information should satisfy, and C ═ is (C ═ C)₁，C₂，...，C_j，...C_m) Representing a set of history records, A ═ A₁，A₂，...，A_j，...A_n) Representing a collection of all attributes, A (C)_i)＝(A₁，A₂，...，A_j，...A_k) Represents a history C_iQ ═ Q (Q), a set of all attributes of₁，q₂，...，q_m) Represents a collection of historical record groupings, | A (C)_i) I represents History C_iThe number of all attributes in (1).

4. The method of claim 3, wherein the characteristic record of each group is a record formed by an average value of all the historical records in the group on the same attribute, and calculating the characteristic record of each group comprises the following steps:

S31)

representing a packet q_jWherein all history records are in attribute A_iAverage value of (3) using

Computing a packet q_jAverage of all history recorded on each attribute, v_k(A_i) Attribute A representing History k_iThe value of (a) is,

representing a packet q_jThe collection of all the history records in the history record,

representing a packet q_jThe number of all history records in the database;

S32)

into a packet q_jIs recorded on the basis of the characteristics of the image,

5. the method of claim 4, wherein calculating the matching value of each of the feature records to the query request is: <math> <mrow> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>input</mi> </msub> <mo>,</mo> <msubsup> <mi>C</mi> <msub> <mi>q</mi> <mi>j</mi> </msub> <mi>r</mi> </msubsup> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <mi>w</mi> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>*</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>input</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> <msub> <mi>v</mi> <msubsup> <mi>C</mi> <msub> <mi>q</mi> <mi>j</mi> </msub> <mi>r</mi> </msubsup> </msub> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> wherein $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{C_{q_j}^r}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{C_{q_j}^r}\left(A_i\right)|}^2,$ C_inputRepresents the request for the query in question,

representing a packet q_jThe characteristic of (2) is recorded.

6. Such asThe method of claim 4, wherein calculating the match value for each history record in the group to which the query request and the feature record having the highest match value belong is: <math> <mrow> <mrow> <mo>(</mo> <msub> <mi>C</mi> <mi>input</mi> </msub> <mo>,</mo> <msub> <mi>C</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <mi>w</mi> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>*</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>input</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> <msub> <mi>v</mi> <msub> <mi>C</mi> <mi>j</mi> </msub> </msub> <mrow> <mo>(</mo> <msub> <mi>A</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> wherein $\mathrm{sim}(v_{\mathrm{input}}\left(A_i\right),v_{c_j}\left(A_i\right))={|v_{\mathrm{input}}\left(A_i\right)-v_{c_j}\left(A_i\right)|}^2,$ C_jRepresenting the history in the group to which the feature record with the highest matching value belongs, C_inputRepresenting the query request.

7. A fabric retrieval system comprises a client and a server, and is characterized in that,

the client comprises:

a fabric database: the system is used for storing fabric-related service data;

the multidimensional data management module: the fabric business management system is used for receiving a query request input by a user and sending the request to a server side, managing data in a fabric database by using a star mode and providing operation related to fabric business;

the server side includes:

the retrieval module: the multidimensional data management module is used for calculating a final weight value of each attribute of the fabric to be retrieved in the query request sent by the multidimensional data management module, grouping the historical records in the historical record database, calculating the characteristic record of each group, calculating the matching value of each characteristic record and the query request, determining the group of the characteristic record with the highest matching value with the query request, calculating the matching value of each historical record and the query request in the group of the characteristic record with the highest matching value, determining the historical record with the highest matching value with the query request, and sending the historical record to the client;

history database: for storing history records.

8. The system of claim 7, wherein the retrieving module comprises:

history grouping unit: the system is used for grouping the historical records in the historical record database and calculating a characteristic record representing each group;

an attribute weight calculation unit: the method comprises the steps of calculating a final weight value of each attribute of the fabric to be retrieved in the query request;

a group search unit: the attribute weight calculation unit is used for calculating a matching value of the query request and each feature record according to the final weight value of each attribute of the fabric to be retrieved in the query request, the value of each attribute in the query request and the value of each attribute of the feature record calculated by the history record grouping unit, and determining the group to which the feature record with the highest matching value with the query request belongs;

history search unit: the attribute weight calculation unit is used for calculating the matching value of the query request and each history record in the group to which the feature record with the highest matching value determined by the group search unit belongs according to the final weight value of each attribute of the fabric to be retrieved in the query request, the value of each attribute in the query request and the value of the attribute of each history record in the group to which the feature record with the highest matching value determined by the group search unit belongs, and determining the history record with the highest matching value with the query request.

Images