CN107748754B - Knowledge graph perfecting method and device - Google Patents
Knowledge graph perfecting method and device Download PDFInfo
- Publication number
- CN107748754B CN107748754B CN201710833203.2A CN201710833203A CN107748754B CN 107748754 B CN107748754 B CN 107748754B CN 201710833203 A CN201710833203 A CN 201710833203A CN 107748754 B CN107748754 B CN 107748754B
- Authority
- CN
- China
- Prior art keywords
- commodity
- knowledge graph
- extracted
- label
- tags
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a knowledge graph perfecting method and a knowledge graph perfecting device, and belongs to the technical field of electronic commerce. The method comprises the following steps: extracting a label of a commodity from commodity information at least comprising a commodity picture; adding the extracted labels to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph; and setting the relation corresponding to the label in the knowledge graph. The embodiment of the invention can enrich the labels in the knowledge map so as to realize the accurate search of the user on the commodities and the accurate recommendation of the user on the purchased commodities.
Description
Technical Field
The invention relates to the technical field of electronic commerce, in particular to a knowledge graph perfecting method and device.
Background
The knowledge graph is a data structure based on a graph, nodes of the data structure represent entities (entries) or concepts (concepts), edges of the data structure represent various semantic relations between the entities/concepts, and complex associated information can be better inquired in a search engine by using the knowledge graph, so that the intention of a user is understood from a semantic level, and the search quality is improved. For example, in the field of electronic commerce, the use of a commodity knowledge graph can help a user to search for a corresponding commodity, and the use of a user knowledge graph can help a user to recommend a commodity.
In the prior art, the commodity knowledge graph or the label of the commodity included in the user knowledge graph generally only identifies basic characteristics such as brand, color, material, size, basic style and the like of the commodity, and the label is directly identified on the commodity after manual review according to information provided by a seller user when uploading the commodity, however, the attention angle of the user to the commodity is more and more at present, if the user wants to search for the commodity according to characteristics other than the basic characteristics of the commodity, for example, searching for a certain commodity with a specified pattern, the knowledge graph does not include the corresponding label, so that the accurate search of the commodity cannot be realized; in addition, accurate recommendation of the user to purchase the commodity cannot be realized from the perspective of the user concerning the commodity.
Therefore, in the prior art, the problem that the commodity labels of the knowledge graph are simple and rough exists, and the knowledge graph needs to be improved to enrich the labels in the knowledge graph so as to realize accurate search of the commodities by the user and accurate recommendation of the commodities purchased by the user.
Disclosure of Invention
In order to solve the problems in the prior art, embodiments of the present invention provide a method and an apparatus for improving a knowledge graph, which enrich tags in the knowledge graph to implement accurate search of a user on a commodity and implement accurate recommendation of a user on a commodity purchased by the user. The technical scheme is as follows:
in a first aspect, a method for improving a knowledge graph is provided, the method comprising:
extracting a label of a commodity from commodity information at least comprising a commodity picture;
adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph;
and setting a relation corresponding to the extracted label in the knowledge graph.
With reference to the first aspect, in a first possible implementation manner, the article information only includes the article picture, and the extracting a label of an article from the article information including the article picture includes:
extracting image features from the commodity picture;
and acquiring a label corresponding to the image characteristic.
With reference to the first aspect, in a second possible implementation manner, the article information includes the article picture and text description information of the article, and the extracting a label of the article from the article information at least including the article picture includes:
extracting image features in the commodity picture; and
extracting key words in the text description information;
determining the label of the commodity according to the image feature and the keyword;
the text description information includes at least one of title information of the commodity, webpage information of the commodity, and comment information of the commodity.
With reference to the first or second possible implementation manner of the first aspect, in a third possible implementation manner,
the adding the extracted tags to the knowledge-graph comprises:
determining the label category corresponding to the extracted label;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting of the relation corresponding to the extracted tag in the knowledge-graph comprises:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
matching the extracted labels with the relationships between other labels in the user knowledge graph; and
updating the extracted relationships between the tags and other tags in the user knowledge graph.
With reference to the first or second possible implementation manner of the first aspect, in a fourth possible implementation manner,
the adding the extracted tags to the knowledge-graph comprises:
adding the extracted tags to the tag library;
defining the label category to which the extracted label belongs;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting of the relation corresponding to the extracted tag in the knowledge-graph comprises:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
updating the extracted relationships between the tags and other tags in the user knowledge graph.
In a second aspect, there is provided a knowledge-graph perfecting apparatus, the apparatus comprising:
the extraction module is used for extracting the label of the commodity from the commodity information at least comprising the commodity picture;
the adding module is used for adding the extracted labels to a knowledge graph, and the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph;
and the setting module is used for setting the relation corresponding to the extracted label in the knowledge graph.
With reference to the second aspect, in a first possible implementation manner, the commodity information only includes the commodity picture, and the extraction module is specifically configured to:
extracting image features from the commodity picture;
and acquiring a label corresponding to the image characteristic.
With reference to the second aspect, in a second possible implementation manner, the article information includes the article picture and text description information of the article, and the extraction module is further specifically configured to:
extracting image features in the commodity picture; and
extracting key words in the text description information;
determining the label of the commodity according to the image feature and the keyword;
the text description information includes at least one of title information of the commodity, webpage information of the commodity, and comment information of the commodity.
With reference to the first or second possible implementation manner of the second aspect, in a third possible implementation manner,
the adding module is specifically configured to:
determining the label category corresponding to the extracted label;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting module is specifically configured to:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
matching the extracted labels with the relationships between other labels in the user knowledge graph; and
updating the extracted relationships between the tags and other tags in the user knowledge graph.
With reference to the first or second possible implementation manner of the second aspect, in a fourth possible implementation manner,
the adding module is specifically further configured to:
adding the tag to the library of tags;
defining a label category to which the label belongs;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting module is specifically further configured to:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
updating the extracted relationships between the tags and other tags in the user knowledge graph.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
by extracting the labels of the commodities from the commodity information at least comprising the commodity pictures and adding the labels to the knowledge graph, wherein the knowledge graph comprises the commodity knowledge graph and/or the user knowledge graph, and setting the relation corresponding to the extracted labels in the knowledge graph, compared with the knowledge graph in the prior art, the knowledge graph only identifies the basic characteristics of the commodities such as brand, color, material, size, basic style and the like, the labels of the commodities extracted from the commodity information at least comprising the commodity pictures can reflect more information and characteristics of the commodities, so that the labels in the commodity knowledge graph and/or the user knowledge graph are richer and more diversified, and the accurate search of the commodities by the user and the accurate recommendation of the commodities purchased by the user are realized.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for improving a knowledge graph according to an embodiment of the present invention;
FIG. 2 is a flow chart of a method for improving a knowledge graph according to an embodiment of the present invention;
FIG. 3 is a flow chart of a method for improving a knowledge graph according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a knowledge-map perfecting apparatus provided in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
The embodiment of the invention provides a knowledge graph perfecting method, which is shown in a reference figure 1 and comprises the following steps:
101. a label of a commodity is extracted from commodity information at least including a commodity picture.
Specifically, the commodity information only includes a commodity picture, and the tag of the commodity is extracted from the commodity information including the commodity picture, and the process may include:
extracting image features from the commodity picture;
labels corresponding to the image features are obtained.
The commodity information comprises a commodity picture and text description information of the commodity, and a label of the commodity is extracted from the commodity information at least comprising the commodity picture, and the process can include:
extracting image features in the commodity picture; and
extracting key words in the text description information;
determining a label of the commodity according to the image characteristics and the keywords;
the text description information includes at least one of title information of the commodity, web page information of the commodity, and comment information of the commodity.
102. And adding the extracted labels to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph.
103. And setting a relation corresponding to the extracted label in the knowledge graph.
The embodiment of the invention provides a knowledge graph perfecting method, which extracts the label of a commodity from commodity information at least comprising commodity pictures, and adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph, and setting a relation corresponding to the extracted tag in the knowledge graph, wherein the tag of the commodity extracted from the commodity information at least comprising the commodity picture can reflect more information and characteristics of the commodity compared with the knowledge graph in the prior art which only identifies basic characteristics such as brand, color, material, size, basic style and the like of the commodity, therefore, the labels in the commodity knowledge map and/or the user knowledge map are richer, more diversified and more refined, therefore, the user can accurately search the commodities and accurately recommend the commodities purchased by the user.
Example two
The embodiment of the invention provides a knowledge graph perfecting method, which is shown in a reference figure 2 and comprises the following steps:
201. a label of a commodity is extracted from commodity information at least including a commodity picture.
The commodity in the commodity picture can be clothing, food, cosmetics, furniture, articles for daily use or other commodities.
The label of the commodity is used for describing the commodity, and the label can identify the name, brand, size, color, style, version and pattern of the commodity or be used for indicating the definition of the generalized or summarized characteristic range in the commodity, and the like, wherein the style can comprise original wind, military travel wind, antique wind, sweet and lovely weather and the like; for indicating a definition of a generalized or summarized range of features in the article, such as "thin", "bright", etc.
Besides, the label may also be other information that can be used to identify the goods, and the embodiment of the present invention does not limit the specific label.
Specifically, the commodity information only includes a commodity picture, and the tag of the commodity is extracted from the commodity information including the commodity picture, and the process may include:
acquiring commodity information at least containing commodity pictures uploaded by a buyer user;
extracting image features from the commodity picture;
labels corresponding to the image features are obtained.
Wherein, the image features are extracted from the commodity picture, and the process may include:
specifically, the process may include:
preprocessing a commodity picture;
and extracting image features in the preprocessed commodity picture.
In this embodiment, the commodity picture is preprocessed to eliminate irrelevant information in the commodity picture, such as filtering interference and noise, and recovering useful real information, so that the reliability of extracting the image features in the commodity picture can be ensured.
Wherein, a label corresponding to the image feature is obtained, the process may include:
acquiring a trained recognition model, wherein the input of the trained recognition model is an image feature, and the output of the trained recognition model is a label corresponding to the image feature, wherein the recognition model can be a deep convolutional neural network;
and when the image features are input into the trained recognition model, determining the labels output by the trained recognition model as labels corresponding to the image features.
For example, the tags that correspond to the image features of the product may include "women's coat", "short money", "academy style", and "grid pattern"; for another example, the labels for acquiring the image feature corresponding to the product may include "satchel", "light brown", "fine", "cowhide", or "satchel", "dark brown", "coarse", "cowhide", and the like.
The label of the commodity is extracted from the commodity information containing the commodity picture, so that more information and characteristics of the commodity can be reflected, and the label of the commodity is rich and detailed.
Specifically, the commodity information includes a commodity picture and text description information of the commodity, and the label of the commodity is extracted from the commodity information at least including the commodity picture, and the process may include:
extracting image features in the commodity picture; and
extracting key words in the text description information;
determining a label of the commodity according to the image characteristics and the keywords;
the text description information of the commodity comprises at least one of title information of the commodity, webpage information of the commodity or comment information of the commodity.
Wherein, extracting the keywords in the text description information, the process may include:
extracting effective texts from the text description information;
performing Chinese word segmentation and part-of-speech tagging on a text, and clustering the text by using an LDA (Latent Dirichlet allocation) model or a PLSA (Latent Semantic content Semantic analysis) model;
after sorting clustering results, carrying out category labeling, carrying out supervised learning, and carrying out classification training on future unlabeled texts;
filtering stop words, and extracting keywords from the clustered texts by using a TF-LDF (term frequency-inverse document frequency) or a TextRank algorithm.
Wherein, according to the image features and the keywords, the label of the goods is determined, and the process may include:
determining a plurality of labels corresponding to the image features and determining a plurality of labels corresponding to the keywords;
carrying out merging and de-duplication processing on the labels corresponding to the image features and the labels corresponding to the keywords;
and determining the tag after the de-duplication processing as a tag extracted from the commodity information input by the user.
Illustratively, the labels corresponding to the image features of the merchandise include "wedding dress", "white", "dress", "gauze", "frizzle pattern", and the labels corresponding to the keywords included in the textual description information of the merchandise include "shoulder tie", "white", "wedding dress", and the labels are subjected to a merge-and-deduplication process, and the labels of the merchandise include "wedding dress", "shoulder tie", "white", "frizzle pattern", "white", and "dress".
In the embodiment of the invention, the label of the commodity is extracted from the commodity information containing the commodity picture and the text description information, so that more information and characteristics of the commodity can be reflected, and the label of the commodity is further enriched, diversified and refined.
202. And determining the label category corresponding to the extracted label.
The knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph, and the label type is at least one or more of the name, the brand, the color, the style, the type and the pattern.
The commodity knowledge graph comprises commodities, a plurality of labels corresponding to the commodities and relationship information among the labels; the user knowledge graph comprises a user, a plurality of labels corresponding to the user and relationship information among the labels.
Specifically, when the extracted labels are identified to be contained in a label library of the knowledge graph, the label categories corresponding to the extracted labels are determined, wherein the label library is provided with the corresponding relations between the labels and the label categories.
Illustratively, the extracted labels are "academy wind" and "beige", the label category corresponding to "academy wind" is "style" and the label category corresponding to "beige" is "color" according to the correspondence between the labels and the label categories.
The present invention does not limit the specific determination process.
203. And adding the extracted labels to the knowledge graph according to the label categories.
Specifically, the present invention does not limit the specific addition process.
It should be noted that, steps 202 to 203 are to implement a process of adding the extracted tag to the knowledge graph, where the knowledge graph includes a commodity knowledge graph and/or a user knowledge graph, and the process may be implemented in other ways besides the above-mentioned steps, and the embodiment of the present invention is not limited in particular ways.
It should be noted that, for adding the extracted tag to the product knowledge graph, the tag may be extracted from the product information at least including the product picture uploaded by the buyer user, and may also be extracted from the product information at least including the product picture uploaded by the seller user.
In the embodiment of the invention, the extracted labels are added to the knowledge graph, so that the improvement of the knowledge graph is realized, and the labels in the knowledge graph are richer, diversified and refined.
204. And setting the relation between the extracted label and the commodity in the commodity knowledge graph.
Specifically, according to the label category corresponding to the extracted label, establishing an association relation between the extracted label and the corresponding commodity in the commodity knowledge map; and
and establishing the relation between the commodity corresponding to the extracted label and other commodities according to the preset commodity collocation information.
In the embodiment of the invention, the extracted labels are added to the commodity knowledge graph, and the corresponding relation with the labels is set in the commodity knowledge graph, so that a user can search corresponding commodities according to the labels in the commodity knowledge graph, and the accurate search of the user on the commodities is realized.
205. The relationships between the extracted tags and other tags are matched in the user knowledge graph.
Specifically, the process may include:
acquiring a matching relation between labels which is established in advance, wherein the matching relation is established based on the relevance between the labels;
and matching the extracted labels with other labels of the user in the user knowledge graph according to the matching relation among the labels.
For example, a matching relationship between the label "sweater" and the label "casual sports" is pre-established, and if the extracted label is "sweater" and the label "casual sports" exists in the user knowledge map, the label "sweater" and the label "casual sports" are matched in the user knowledge map.
206. And updating the relation between the extracted label and other labels in the user knowledge graph.
Specifically, the process may include:
acquiring historical purchase records of a user, and determining an interest model of the user according to the historical purchase records, wherein the interest model is at least used for indicating at least one of consumption preference, preference description and collocation preference of the user;
the relationships between the tags and tags belonging to other tag categories are updated in the user's knowledge graph according to the user's interest model.
The relation between the extracted label and other labels is updated in the user knowledge graph, the favorite interest of the user is further described, and therefore accurate recommendation of the user for purchasing commodities is further achieved.
It should be noted that steps 204 to 206 are processes for implementing setting of the relationship corresponding to the tag in the knowledge graph, and the processes may be implemented in other ways besides the above-mentioned ways, and the specific way is not limited by the embodiment of the present invention.
It should be noted that, in the embodiment of the present invention, the execution sequence of step 204, step 205, and step 206 is specifically defined, and in practical application, step 204, step 205, and step 206 are executed at the same time, which is a preferable scheme, so as to improve the efficiency of improving the knowledge graph.
In the embodiment of the invention, the extracted labels are added to the user knowledge graph, and the corresponding relation with the labels is set in the user knowledge graph, so that the labels related to the user in the user knowledge graph are richer, diversified and refined, and the accurate recommendation of the user for purchasing commodities is realized.
The embodiment of the invention provides a knowledge graph perfecting method, which extracts the label of a commodity from commodity information at least comprising commodity pictures, and adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph, and setting a relation corresponding to the extracted tag in the knowledge graph, wherein the tag of the commodity extracted from the commodity information at least comprising the commodity picture can reflect more information and characteristics of the commodity compared with the knowledge graph in the prior art which only identifies basic characteristics such as brand, color, material, size, basic style and the like of the commodity, therefore, the labels in the commodity knowledge graph and/or the user knowledge graph are richer and more diversified, therefore, the user can accurately search the commodities and accurately recommend the commodities purchased by the user.
EXAMPLE III
The embodiment of the invention provides a knowledge graph perfecting method, which is shown in a reference figure 3 and comprises the following steps:
301. a label of a commodity is extracted from commodity information at least including a commodity picture.
Specifically, the step is the same as step 201, and is not described herein again.
In the embodiment of the invention, the label of the commodity is extracted from the commodity information containing the commodity picture and the text description information, so that more information and characteristics of the commodity can be reflected, and the label of the commodity is further enriched, diversified and refined.
302. Adding the extracted tags to a tag library.
Specifically, when the extracted tags are identified not to be included in the tag library of the knowledge graph, the extracted tags are added to the tag library.
The present invention does not limit the specific determination process.
The extracted tags are added into the tag library which does not contain the tags originally, so that the tags contained in the tag library are richer.
303. And defining the label category to which the extracted label belongs.
Specifically, whether the label library has the label with the same attribute as the extracted label is judged;
if yes, determining the label type corresponding to the label with the same attribute as the extracted label as the label type corresponding to the extracted label;
otherwise, a label category corresponding to the mentioned label is created.
304. And adding the extracted labels to the knowledge graph according to the label categories.
The present invention is not limited to the specific definition process.
It should be noted that steps 302 to 304 are processes for implementing adding extracted labels to the knowledge graph, and the processes may be implemented in other ways besides the above-mentioned steps, and the specific way is not limited by the embodiment of the present invention.
It should be noted that, for adding the extracted tag to the product knowledge graph, the tag may be extracted from the product information at least including the product picture uploaded by the buyer user, and may also be extracted from the product information at least including the product picture uploaded by the seller user.
In the embodiment of the invention, the extracted labels are added to the corresponding label categories, so that the improvement of the knowledge graph is realized, and the labels in the knowledge graph are richer, diversified and refined.
305. And setting the relation between the extracted label and the commodity in the commodity knowledge graph.
Specifically, the step is the same as step 204, and is not described herein again.
In the embodiment of the invention, the extracted labels are added to the commodity knowledge graph, and the corresponding relation with the labels is set in the commodity knowledge graph, so that a user can search corresponding commodities according to the labels in the commodity knowledge graph, and the accurate search of the user on the commodities is realized.
306. And updating the relation between the extracted label and other labels of the user in the knowledge graph of the user.
Specifically, the step is the same as step 206, and is not described herein again.
It should be noted that steps 304 to 305 are processes for implementing setting of the relationship corresponding to the extracted tag in the knowledge graph, and the processes may be implemented in other ways besides the above-mentioned ways, and the embodiment of the present invention does not limit the concrete ways.
It should be noted that, in the embodiment of the present invention, the execution sequence of the step 304 and the step 305 is specifically defined, and in practical application, the step 304 and the step 305 are executed at the same time, which is a preferable scheme, so as to improve the efficiency of improving the knowledge graph.
In the embodiment of the invention, the extracted labels are added to the user knowledge graph, and the relation corresponding to the extracted labels is set in the user knowledge graph, so that the labels related to the user in the user knowledge graph are richer, diversified and refined, and the accurate recommendation of the user for purchasing commodities is realized.
The embodiment of the invention provides a knowledge graph perfecting method, which extracts the label of a commodity from commodity information at least comprising commodity pictures, and adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph, and setting a relation corresponding to the extracted tag in the knowledge graph, wherein the tag of the commodity extracted from the commodity information at least comprising the commodity picture can reflect more information and characteristics of the commodity compared with the knowledge graph in the prior art which only identifies basic characteristics such as brand, color, material, size, basic style and the like of the commodity, therefore, the labels in the commodity knowledge graph and/or the user knowledge graph are richer and more diversified, therefore, the user can accurately search the commodities and accurately recommend the commodities purchased by the user.
Example four
The embodiment of the invention provides a knowledge graph perfecting device, and as shown in fig. 4, the device 4 comprises:
an extracting module 41, configured to extract a tag of a product from product information at least including a product picture;
an adding module 42, configured to add the extracted tag to a knowledge graph, where the knowledge graph includes a commodity knowledge graph and/or a user knowledge graph;
a setting module 43, configured to set a relationship corresponding to the extracted tag in the knowledge graph.
Optionally, the commodity information only includes a commodity picture, and the extraction module 41 is specifically configured to:
extracting image features from the commodity picture;
labels corresponding to the image features are obtained.
Optionally, the commodity information includes a commodity picture and text description information of the commodity, and the extraction module 41 is further specifically configured to:
extracting image features in the commodity picture; and
extracting key words in the text description information;
determining a label of the commodity according to the image characteristics and the keywords;
the text description information includes at least one of title information of the commodity, web page information of the commodity, and comment information of the commodity.
Optionally, the adding module 42 is specifically configured to:
determining the label category corresponding to the extracted label;
adding the extracted labels to the knowledge graph according to the label categories;
the setting module 43 is specifically configured to:
setting the relation between the extracted labels and the commodities in a commodity knowledge map;
matching the extracted labels with the relationships between other labels in the user knowledge graph; and
and updating the relation between the extracted label and other labels in the user knowledge graph.
Optionally, the adding module 42 is further specifically configured to:
adding tags to a library of tags;
defining the label category to which the label belongs;
adding the extracted labels to the knowledge graph according to the label categories;
the setting module 43 is specifically further configured to:
setting the relation between the extracted labels and the commodities in a commodity knowledge map;
and updating the relation between the extracted label and other labels in the user knowledge graph.
The embodiment of the invention provides a knowledge graph perfecting device, which extracts the label of a commodity from commodity information at least comprising commodity pictures, and adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph, and setting a relation corresponding to the extracted tag in the knowledge graph, wherein the tag of the commodity extracted from the commodity information at least comprising the commodity picture can reflect more information and characteristics of the commodity compared with the knowledge graph in the prior art which only identifies basic characteristics such as brand, color, material, size, basic style and the like of the commodity, therefore, the labels in the commodity knowledge graph and/or the user knowledge graph are richer and more diversified, therefore, the user can accurately search the commodities and accurately recommend the commodities purchased by the user.
All the above-mentioned optional technical solutions can be combined arbitrarily to form the optional embodiments of the present invention, and are not described herein again.
It should be noted that: in the knowledge-graph improvement device provided in the above embodiment, when the knowledge-graph improvement method is executed, only the division of the functional modules is illustrated, and in practical application, the function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the knowledge graph improvement device provided by the above embodiment and the knowledge graph improvement method embodiment belong to the same concept, and the specific implementation process thereof is described in the method embodiment and is not described herein again.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by associated hardware through a program, and the program may be stored in a computer readable storage medium, and the above mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. A method of knowledge-graph improvement, the method comprising:
the method for extracting the label of the commodity from the commodity information at least comprising the commodity picture and the text description information of the commodity comprises the following steps: extracting image features in the commodity picture, extracting effective texts from text description information, performing Chinese word segmentation and part-of-speech tagging on the effective texts, clustering the texts by using an LDA (latent Dirichlet Allocation) model or a PLSA (partial least squares) model, extracting key words in the clustered texts by using a TF-LDF (Trans-Linear discriminant function) algorithm or a TextRank algorithm, and determining labels of the commodities according to the image features and the key words, wherein the text description information comprises at least one of title information of the commodities, webpage information of the commodities and comment information of the commodities;
adding the extracted tags to a knowledge graph, wherein the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph;
and setting a relation corresponding to the extracted label in the knowledge graph.
2. The method of claim 1,
the adding the extracted tags to the knowledge-graph comprises:
determining the label category corresponding to the extracted label;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting of the relation corresponding to the extracted tag in the knowledge-graph comprises:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
matching the extracted labels with the relationships between other labels in the user knowledge graph; and
updating the extracted relationships between the tags and other tags in the user knowledge graph.
3. The method of claim 1,
the adding the extracted tags to the knowledge-graph comprises:
adding the extracted tags to a tag library;
defining the label category to which the extracted label belongs;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting of the relation corresponding to the extracted tag in the knowledge-graph comprises:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
updating the extracted relationships between the tags and other tags in the user knowledge graph.
4. A knowledge-graph perfecting apparatus, said apparatus comprising:
the extraction module is used for extracting the label of the commodity from the commodity information at least comprising the commodity picture and the text description information of the commodity, and comprises the following steps: extracting image features in the commodity picture, extracting effective texts from text description information, performing Chinese word segmentation and part-of-speech tagging on the effective texts, clustering the texts by using an LDA (latent Dirichlet Allocation) model or a PLSA (partial least squares) model, extracting key words in the clustered texts by using a TF-LDF (Trans-Linear discriminant function) algorithm or a TextRank algorithm, and determining labels of the commodities according to the image features and the key words, wherein the text description information comprises at least one of title information of the commodities, webpage information of the commodities and comment information of the commodities; the adding module is used for adding the extracted labels to a knowledge graph, and the knowledge graph comprises a commodity knowledge graph and/or a user knowledge graph;
and the setting module is used for setting the relation corresponding to the extracted label in the knowledge graph.
5. The apparatus of claim 4,
the adding module is specifically configured to:
determining the label category corresponding to the extracted label;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting module is specifically configured to:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
matching the extracted labels with the relationships between other labels in the user knowledge graph; and
updating the extracted relationships between the tags and other tags in the user knowledge graph.
6. The apparatus of claim 4,
the adding module is specifically further configured to:
adding the tag to the library of tags;
defining a label category to which the label belongs;
adding the extracted tags to the knowledge graph according to the tag categories;
the setting module is specifically further configured to:
setting the relation between the extracted label and the commodity in the commodity knowledge graph;
updating the extracted relationships between the tags and other tags in the user knowledge graph.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710833203.2A CN107748754B (en) | 2017-09-15 | 2017-09-15 | Knowledge graph perfecting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710833203.2A CN107748754B (en) | 2017-09-15 | 2017-09-15 | Knowledge graph perfecting method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107748754A CN107748754A (en) | 2018-03-02 |
| CN107748754B true CN107748754B (en) | 2021-10-26 |
Family
ID=61255746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710833203.2A Active CN107748754B (en) | 2017-09-15 | 2017-09-15 | Knowledge graph perfecting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107748754B (en) |
Families Citing this family (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108427749B (en) * | 2018-03-12 | 2021-06-18 | 上海哔哩哔哩科技有限公司 | Commodity label management method, server and commodity label management system |
| CN108710695B (en) * | 2018-05-23 | 2019-08-06 | 掌阅科技股份有限公司 | Mind map generation method and electronic equipment based on e-book |
| CN109145098B (en) * | 2018-07-20 | 2021-10-29 | 西北大学 | Chinese culture element information searching method based on knowledge graph |
| CN109189937B (en) | 2018-08-22 | 2021-02-09 | 创新先进技术有限公司 | Feature relationship recommendation method and device, computing device and storage medium |
| CN109189938A (en) * | 2018-08-31 | 2019-01-11 | 北京字节跳动网络技术有限公司 | Method and apparatus for updating knowledge mapping |
| CN109299287A (en) * | 2018-10-24 | 2019-02-01 | 深圳素问智能信息技术有限公司 | A kind of querying method and device of drinks information |
| CN109710776B (en) * | 2018-12-29 | 2022-10-28 | 中国科学技术大学 | Construction method of knowledge graph of photo album |
| CN109840288A (en) * | 2019-02-21 | 2019-06-04 | 北京一品智尚信息科技有限公司 | Picture retrieval method, equipment and storage medium |
| CN112784063A (en) * | 2019-03-15 | 2021-05-11 | 北京金山数字娱乐科技有限公司 | Idiom knowledge graph construction method and device |
| CN110609903B (en) * | 2019-08-01 | 2022-11-11 | 华为技术有限公司 | Information presentation method and device |
| CN110737779B (en) * | 2019-09-18 | 2021-05-04 | 北京三快在线科技有限公司 | Knowledge graph construction method and device, storage medium and electronic equipment |
| CN110990584B (en) * | 2019-11-26 | 2021-02-09 | 口口相传(北京)网络技术有限公司 | Knowledge graph generation method and device |
| CN111143575B (en) * | 2019-12-06 | 2023-07-04 | 苏宁云计算有限公司 | Optimization method and system for store commodity structure |
| CN111738802A (en) * | 2020-06-30 | 2020-10-02 | 广东奥园奥买家电子商务有限公司 | Introduction method, device and equipment of E-commerce commodities |
| CN112102029B (en) * | 2020-08-20 | 2022-10-14 | 浙江大学 | Knowledge graph-based long-tail recommendation calculation method |
| CN113535985A (en) * | 2021-08-17 | 2021-10-22 | 北京爱奇艺科技有限公司 | Object label determination method and device |
| CN113505245A (en) * | 2021-09-10 | 2021-10-15 | 深圳平安综合金融服务有限公司 | Knowledge graph generation method, computer readable storage medium and computer device |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5241671C1 (en) * | 1989-10-26 | 2002-07-02 | Encyclopaedia Britannica Educa | Multimedia search system using a plurality of entry path means which indicate interrelatedness of information |
| US7917492B2 (en) * | 2007-09-21 | 2011-03-29 | Limelight Networks, Inc. | Method and subsystem for information acquisition and aggregation to facilitate ontology and language-model generation within a content-search-service system |
| CN102254043B (en) * | 2011-08-17 | 2013-04-03 | 电子科技大学 | Semantic mapping-based clothing image retrieving method |
| CN104750795B (en) * | 2015-03-12 | 2017-09-01 | 北京云知声信息技术有限公司 | A kind of intelligent semantic searching system and method |
| CN105550190B (en) * | 2015-06-26 | 2019-03-29 | 许昌学院 | Cross-media retrieval system towards knowledge mapping |
| CN106355627A (en) * | 2015-07-16 | 2017-01-25 | 中国石油化工股份有限公司 | Method and system used for generating knowledge graphs |
| CN105912656B (en) * | 2016-04-07 | 2020-03-17 | 桂林电子科技大学 | Method for constructing commodity knowledge graph |
| CN106708940B (en) * | 2016-11-11 | 2020-06-30 | 百度在线网络技术（北京）有限公司 | Method and device for processing pictures |
| CN106815307A (en) * | 2016-12-16 | 2017-06-09 | 中国科学院自动化研究所 | Public Culture knowledge mapping platform and its use method |
-
2017
- 2017-09-15 CN CN201710833203.2A patent/CN107748754B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN107748754A (en) | 2018-03-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107748754B (en) | Knowledge graph perfecting method and device | |
| CN107833082B (en) | Commodity picture recommendation method and device | |
| CN107705066B (en) | Information input method and electronic equipment during commodity warehousing | |
| KR102075833B1 (en) | Curation method and system for recommending of art contents | |
| CN110263180B (en) | Intention knowledge graph generation method, intention identification method and device | |
| CN110059271B (en) | Searching method and device applying tag knowledge network | |
| TWI631474B (en) | Method and device for product identification label and method for product navigation | |
| CN106294425B (en) | The automatic image-text method of abstracting and system of commodity network of relation article | |
| CN104636402B (en) | A kind of classification of business object, search, method for pushing and system | |
| US10360623B2 (en) | Visually generated consumer product presentation | |
| CN105550369B (en) | A kind of method and device for searching for end article collection | |
| CN109815308A (en) | The determination of intention assessment model and retrieval intension recognizing method, device | |
| CN107632984A (en) | A kind of cluster data table shows methods, devices and systems | |
| US20230214895A1 (en) | Methods and systems for product discovery in user generated content | |
| KR20210098884A (en) | A method of providing a fashion item recommendation service using a body shape and purchase history | |
| US20200226168A1 (en) | Methods and systems for optimizing display of user content | |
| CN115048583A (en) | Information pushing method, system, equipment and storage medium based on user portrait | |
| CN107861971A (en) | A kind of product search method and device | |
| CN107908749B (en) | Character retrieval system and method based on search engine | |
| CN113744019A (en) | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium | |
| CN113821718A (en) | Article information pushing method and device | |
| US11410418B2 (en) | Methods and systems for tagged image generation | |
| Lodkaew et al. | Fashion finder: A system for locating online stores on instagram from product images | |
| KR101498944B1 (en) | Method and apparatus for deciding product seller related document | |
| Kang et al. | Recognising informative Web page blocks using visual segmentation for efficient information extraction. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230106 Address after: 510123 building 6, No. 20, Huahai street, Fangcun, Liwan District, Guangzhou, Guangdong (office only) Patentee after: GUANGZHOU PINWEI SOFTWARE Co.,Ltd. Address before: 510000 room 01, No.314, Fangcun Avenue middle, Liwan District, Guangzhou City, Guangdong Province Patentee before: GUANGZHOU WEIPINHUI RESEARCH INSTITUTE CO.,LTD. |
|
| TR01 | Transfer of patent right |