TWI640881B - Data analyzing device and data analyzing method - Google Patents

Abstract

The invention provides a data analysis device and a data analysis method. The data analysis device comprises: an input and output module, and receives input data, wherein the input data includes a product name and a plurality of image words; the search module searches for multiple images and corresponding maps on the plurality of network platforms according to the input data. Image information; identification module, identifying text information for product information and user information; analysis module, obtaining multiple image classifications and image values corresponding to each image based on user information; and feedback module And generating an image correlation map corresponding to the image word according to the image classification and the image value, and transmitting the image correlation map to the input/output module.

Description

Data analysis device and data analysis method

The invention relates to a data analysis device and a data analysis method.

Data collection is a situation that product development will definitely face. Generally speaking, before developing a product, the designer will first read the written materials or use electronic tools to collect relevant product information, so it takes extra time to organize the collected information. This method is time consuming and the information collected is less sourced. Some of the larger companies develop their own product information collection software, but such systems are less open and generally unavailable to the general user. As a result, most product developers can only use existing methods to collect data, which increases the amount of time it takes to collect data. If the product developer does not have much time to collect the data, he can only choose to simplify the process of data collection and analysis, which may make the designed product not meet expectations.

In order to collect information on a specific product, it is often necessary to find information on e-commerce stores, manufacturer websites, and product catalogs of different regions in different regions, which may result in more time spent on competing product analysis and market research. In addition, the traditional data collection method can not directly obtain feedback from customers. Therefore, the manufacturing end often needs to design products through the opinions given by the sales side, which leads the manufacturing end to understand the thoughts and opinions of the demand side. Therefore, how to quickly sort out a large number of consumer reviews of related products to assist the design process of products is a goal that those skilled in the art should strive for.

In view of this, the present invention provides a data analysis device and a data analysis method, which can quickly organize user opinions and graphically display image correlation maps of various related products to provide product designers as a reference or make further decisions. analysis.

The invention provides a data analysis device, comprising: an input and output module, receiving input data, wherein the input data includes a product name and a plurality of image words; the search module is coupled to the input and output module, and the search module is based on the input data. Searching for multiple images and text information corresponding to each image on multiple network platforms; the identification module is coupled to the search module, the identification module recognizes text information to obtain product information and user information; and the analysis module And coupled to the identification module, the analysis module obtains multiple image classifications and multiple image values corresponding to each image according to the user information; and the feedback module is coupled to the analysis module and the input and output module, and feedback The module generates an image correlation map corresponding to the image word according to the image classification and the image value, and transmits the image correlation map to the input and output module.

The present invention provides a data analysis method, including: receiving input data through an input/output module, wherein the input data includes a product name and a plurality of image words; and the search module searches for multiple images on multiple network platforms according to the input data. The image and the text information corresponding to each image; the identification module recognizes the text information to obtain product information and user information; and the analysis module obtains a plurality of image classifications corresponding to each image according to the user information and a plurality of image values; and generating, by the feedback module, the image correlation map corresponding to the image word according to the image classification and the image value, and transmitting the image correlation map to the input and output module.

The above described features and advantages of the invention will be apparent from the following description.

The data analysis device and the data analysis method of the present invention collect product images and user feedback opinions on product names and image words to be developed by product developers on multiple network platforms, and feedback opinions from users. The lieutenant takes out multiple adjectives for the product as an image classification, and then generates an image correlation map according to the image classification result to provide a reference for the product developer. The image correlation diagram is a two-dimensional correlation diagram, and the horizontal axis and the vertical axis respectively represent the image values of the image classification of the related products, and the images of the related products are displayed on the corresponding coordinates, so the product designer can refer to the image. Quickly know the product closest to a specific image value by associating the graph. In addition, the data analysis device and the data analysis method of the present invention also perform semantic expansion on the image words input by the designer to assist in the horizontal design of the design. The following description will be made by way of examples.

1 is a block diagram of a data analysis apparatus in accordance with an embodiment of the present invention.

Referring to FIG. 1 , the data analysis apparatus 100 of the present invention includes an input/output module 110 , a search module 130 , an identification module 150 , an analysis module 170 , and a feedback module 190 . The search module 130 is coupled to the input and output module 110. The identification module 150 is coupled to the search module 130. The analysis module 170 is coupled to the identification module 150. The feedback module 190 is coupled to the analysis module 170 and the input and output module 110. The data analysis device 100 further includes a processor (not shown) and a memory (not shown). The processor can be used to execute related functions of the input/output module 110, the search module 130, the identification module 150, the analysis module 170, and the feedback module 190, and the memory can store or temporarily store the input/output module 110 and the search module. The data that is present, received, or generated in the group 130, the identification module 150, the analysis module 170, and the feedback module 190.

The processor can be a Central Processing Unit (CPU), or other programmable general purpose or special purpose microprocessor (Microprocessor), digital signal processor (DSP), and can be programmed. Controller, Application Specific Integrated Circuit (ASIC) or other similar components or a combination of the above.

The memory can be any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk ( Hard Disk Drive (HDD), solid state drive (SSD) or similar components or a combination of the above.

Although the modules in this embodiment are disposed in the same data analysis device 100, the present invention is not limited thereto. In another embodiment, the modules may be partially or completely disposed in different devices, and communicate with other modules through a communication chip in a wired or wireless manner.

The input and output module 110 can receive input data of a product designer, and the input data includes a product name and a plurality of image words. For example, the user can enter the product name "scissors" and the image words "useful", "convenience" and the like. The imagery here refers to an adjective corresponding to a product, or an adjective related to the user's expectation of the product. The input/output module 110 can also perform semantic expansion on the image word, and expand the image word into a plurality of synonymous words of similar meaning or similar meaning based on the product name. For example, when the product name is "scissors", "useful" can be expanded to "images" such as "good", "sharp", "ergonomics" and "comfort", and "convenience" can be expanded to "portable". Imagery such as "small".

The search module 130 can search for multiple images and text information corresponding to each image on multiple network platforms according to the input data obtained by the input/output module 110. Specifically, when the user inputs the product name and the image word, the search module 130 (for example, a crawler) can serialize the content on the network platform such as the e-commerce, the open platform, and the web blog according to the product name and the image word. To collect images and product information or related information such as sharing of products related to the product name. In addition to the language of the input data, the search module 130 can also search for related products in other languages and their corresponding information.

The identification module 150 can recognize text information to obtain product information and user information. Specifically, the identification module 150 can extract product information such as product type, product size, and product color from the entire text of the product, and extract user information such as user gender and preferences from the user's mind or User feedback on this product. The feedback information may include multiple adjectives by the user for this product.

The analysis module 170 can obtain a plurality of image classifications and a plurality of image values corresponding to each of the images according to the user information. That is to say, the analysis module 170 can perform image classification of a product or product image according to multiple adjectives in the feedback information. When there are more adjectives in the network platform to describe a product, the product corresponds to this. The image value of the image classification of the adjective will be higher. However, the invention is not limited thereto. In another embodiment, the analysis module 170 can also calculate the image classification of a product and the image value of each image classification through a deep learning method to obtain a more accurate image correlation calculation result.

It is worth noting that the analysis module 170 can also translate user information in multiple languages to obtain image classification and image values corresponding to each image, so as to collect relevant feedback information of the same product by users all over the world. .

In an embodiment, the analysis module 170 can also semantically expand the adjectives in the network platform to obtain an image classification.

The feedback module 190 can generate an image association map corresponding to the image words according to the image classification and the image value, and transmit the image correlation map to the input/output module 110 for providing to the user. Specifically, the feedback module 190 generates one or more two-dimensional image correlation maps, and uses the image classification corresponding to the two image words as the horizontal axis and the vertical axis of the two-dimensional image correlation map, and according to the above two The image value corresponding to the image word displays the image on the two-dimensional image correlation map, as shown in FIG.

In addition, the feedback module 190 can also display the price, the number of comments, the star rating, and the like of each related product in the two-dimensional image association map corresponding to the product image (not shown in the figure). The user can also click on the product image to link to the original material of the product.

Since the image association map is displayed in two dimensions in this embodiment, when the user's input data or input data is semantically expanded to include 2N image words, the feedback module 190 can generate N two-dimensional image associations. The map is transmitted to the input and output module 110 for delivery to the user. For example, a superposition of a first image related map and a second image related graph generated by the input image word or the input image word after the semantic expansion is shown in FIG. When the sharpness and convenience of the product developer's needs are in the scope of the region 310 and the degree of demand and the degree of miniaturization are in the range of the region 320, the product developer can find the region 310 from the overlapping image correlation map. The common products in area 320 are used as a basis for development. Although the above is an example of a stack of two image related maps, the present invention is not limited thereto. In another embodiment, the feedback module 190 can also generate a stack of three or more two-dimensional image related images and transmit them to the input and output module 110 for providing the user with a design reference or direction.

In another embodiment, the input and output module 110 can also receive an input value corresponding to each image word, and extract an image with the smallest difference between the image value and the input value from the plurality of two-dimensional association graphs, directly drawing the image. The image is transmitted to the input/output module 110.

For example, suppose the product name is "scissors" and one of the image words is "sharp". If the product developer wants to develop children's scissors, you can enter the image value corresponding to the image word "sharp", for example between 2.5 and 3.5 between the sharpness range 0 to 10 (as shown in the two dotted lines in Figure 2). Range) to meet the lower sharpness for children. Similarly, the product developer can input an input value corresponding to all other image words, and the image with the smallest difference between the image value and the input value can be directly obtained from the input and output module 110 as a design reference.

In an embodiment, the feedback module 190 can also generate a three-dimensional image association map, using the image classification corresponding to the two image words as the X-axis and the Y-axis of the three-dimensional image correlation map, with the price of the product, the number of comments, or the star The grade data is used as the Z axis of the three-dimensional image correlation map, and the image is displayed on the three-dimensional image correlation map according to the image value corresponding to the two image words and the price, the number of comments or the star rating of the product. For example, in FIG. 4, the X, Y, and Z axes of the three-dimensional image correlation map may represent the sharpness, convenience, and price of the scissors, respectively. For example, the scissors 400 has a sharpness of 7.5, a convenience level of 4, and a price of 450. Through the three-dimensional image correlation diagram, product developers can more clearly understand the relationship between each image classification of the product and the price of the product, the number of comments or the star rating.

Although in the example of FIG. 4, the three axes of the three-dimensional image correlation diagram include the image values corresponding to the two image words and the price of the product, the number of comments, or the star rating, the present invention is not limited thereto. In another embodiment, the three axes of the three-dimensional image association graph may also include image values corresponding to the three image words.

In summary, the data analysis apparatus 100 of the present invention can receive the intuitive condition input of the user, and search for product related data and consumer feedback on the back end, and then perform semantic processing on related data and consumer feedback. The above semantic processing includes word breaking, sentence breaking, semantic analysis and induction. In this way, products with similar semantics can be searched and visually presented on the image association map.

FIG. 5 is a flowchart of a data parsing method according to an embodiment of the invention.

In step S501, input data is received, wherein the input data includes a product name and a plurality of image words.

In step S503, a plurality of images and text information corresponding to each image are searched on a plurality of network platforms according to the input data.

In step S505, the text information is recognized to obtain product information and user information.

In step S507, a plurality of image classifications and a plurality of image values corresponding to each image are obtained based on the user information.

In step S509, an image association map corresponding to the image word is generated according to the image classification and the image value, and the image correlation map is output.

In summary, the data analysis device and the data analysis method of the present invention collect product images and user feedback opinions on product names and image words to be developed by product developers on multiple network platforms, and In the feedback opinion of the user, a plurality of adjectives for the product are taken out as image classification, and the image correlation diagram is generated according to the image classification result to be provided to the product developer for reference. The image correlation diagram is a two-dimensional correlation diagram, and the horizontal axis and the vertical axis respectively represent the image values of the image classification of the related products, and the images of the related products are displayed on the corresponding coordinates, so the product designer can refer to the image. Quickly know the product closest to a specific image value by associating the graph. In addition, the data analysis device and the data analysis method of the present invention also perform semantic expansion on the image words input by the designer to assist in the design level and accelerate the design of the product.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

100‧‧‧Data analysis device

110‧‧‧Input and output modules

130‧‧‧Search Module

150‧‧‧ Identification Module

170‧‧‧Analysis module

190‧‧‧Reward module

310, 320‧‧‧ area

400‧‧‧Scissors

Steps for S501, S503, S505, S507, S509‧‧‧ data analysis methods

1 is a block diagram of a data analysis apparatus in accordance with an embodiment of the present invention. 2 is a schematic diagram of a two-dimensional image correlation diagram in accordance with an embodiment of the present invention. 3 is a schematic diagram of a plurality of two-dimensional image correlation maps stacked in accordance with an embodiment of the present invention. 4 is a schematic diagram of a three-dimensional image correlation diagram in accordance with an embodiment of the present invention. FIG. 5 is a flowchart of a data parsing method according to an embodiment of the invention.

Claims (16)

A data analysis device includes: an input/output module that receives an input data, wherein the input data includes a product name and a plurality of image words; a search module coupled to the input and output module, the search module Searching for a plurality of images on a plurality of network platforms and a text message corresponding to each of the images; the identification module is coupled to the search module, and the recognition module recognizes the text information Obtaining a product information and a user information; an analysis module coupled to the identification module, the analysis module obtaining a plurality of image classifications and a plurality of image values corresponding to each of the images according to the user information The analysis module translates the user information in the plurality of languages to obtain the image classifications and the image values corresponding to each of the images; and a feedback module coupled to the analysis module and the And an input and output module, the feedback module generates an image correlation map corresponding to the image words according to the image classifications and the image values, and transmits the image correlation map to the input and output module. The data analysis device of claim 1, wherein the image correlation map is a two-dimensional association map, and the feedback module uses the image classifications corresponding to two of the image words as the two-dimensional association. A horizontal axis and a vertical axis of the graph, and the images are displayed on the image correlation map according to the image values corresponding to the two image words. The data analysis device of claim 1, wherein the image correlation map is a three-dimensional association map, and the feedback module uses the image classifications corresponding to three of the image words as the three-dimensional association map. Three axes, and displaying the images on the image correlation map according to the image values corresponding to the three image words. The data analysis device of claim 1, wherein the product information includes a product category, a product size, and a product color, and the user information includes a person information and a feedback information, and the analysis module captures The plurality of adjectives in the feedback information obtain the image classifications, and the image values are obtained according to the number of each of the adjectives. The data analysis device of claim 4, wherein the analysis module performs a semantic expansion on the adjectives to obtain the image classifications, wherein the semantic expansion expands each of the adjectives into a plurality of synonyms or An adjective of similar meaning. The data analysis device of claim 1, wherein the input/output module performs a semantic expansion on the image words, and the feedback module generates the correspondence according to the image classifications and the image values. An image association diagram of the expanded image words, wherein the semantic expansion expands each of the image words into a plurality of synonymous or similar meaning image words. The data analysis device of claim 1, wherein the feedback module displays a price, a number of comments, and a star in the product information on the images of the image association map. The data analysis device of claim 1, wherein the input/output module receives a plurality of input values corresponding to the image words, and the feedback module generates corresponding values according to the image classifications and the image values. a plurality of two-dimensional association graphs of the image words, and extracting, from the two-dimensional correlation graphs, a first image of the image values having the smallest difference between the image values and the input values, and the first image The image is transferred to the input and output module. A data analysis method includes: receiving an input data by an input/output module, wherein the input data includes a product name and a plurality of image words; and a search module is configured on the plurality of network platforms according to the input data Searching for a plurality of images and a text message corresponding to each of the images; identifying the text information by an identification module to obtain a product information and a user information; and obtaining, by the analysis module, the user information Corresponding to a plurality of image classifications and a plurality of image values of each of the images, wherein the analysis module translates the user information in the plurality of languages to obtain the image classifications corresponding to each of the images and the imagery And generating, by a feedback module, an image association map corresponding to the image words according to the image classifications and the image values, and transmitting the image correlation map to the input and output module. The data analysis method of claim 9, wherein the image correlation diagram is a two-dimensional association diagram, and the feedback module uses the image classifications corresponding to two of the image words as the two-dimensional association. A horizontal axis and a vertical axis of the graph, and the images are displayed on the image correlation map according to the image values corresponding to the two image words. The data analysis method of claim 9, wherein the image correlation diagram is a three-dimensional association diagram, and the feedback module uses the image classifications corresponding to three of the image words as the three-dimensional association diagram. Three axes, and displaying the images on the image correlation map according to the image values corresponding to the three image words. The data analysis method of claim 9, wherein the product information includes a product type, a product size, and a product color, and the user information includes a person information and a feedback information, and the analysis module captures The plurality of adjectives in the feedback information obtain the image classifications, and the image values are obtained according to the number of each of the adjectives. The data analysis method according to claim 12, wherein the analysis module performs a semantic expansion on the adjectives to obtain the image classifications, wherein the semantic expansion expands each of the adjectives into a plurality of synonyms or An adjective of similar meaning. The data analysis method according to claim 9, wherein the input/output module performs a semantic expansion on the image words, and the feedback module generates the semantics according to the image classifications and the image values. An image association diagram of the expanded image words, wherein the semantic expansion expands each of the image words into a plurality of synonymous or similar meaning image words. The data analysis method of claim 9, wherein the feedback module displays a price, a number of comments, and a star in the product information on the images of the image association map. The data analysis method of claim 9, wherein the input/output module receives a plurality of input values corresponding to the image words, and the feedback module generates corresponding icons according to the image classifications and the image values. a plurality of two-dimensional association graphs of the image words, and extracting, from the two-dimensional correlation graphs, a first image of the image values having the smallest difference between the image values and the input values, and the first image The image is transferred to the input and output module.