CN115017419A - Customized pet food method and system based on personalized recommendation - Google Patents
Customized pet food method and system based on personalized recommendation Download PDFInfo
- Publication number
- CN115017419A CN115017419A CN202210953287.4A CN202210953287A CN115017419A CN 115017419 A CN115017419 A CN 115017419A CN 202210953287 A CN202210953287 A CN 202210953287A CN 115017419 A CN115017419 A CN 115017419A
- Authority
- CN
- China
- Prior art keywords
- pet food
- user
- cluster
- recommendation
- clustering
- 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.)
- Pending
Links
- 235000013305 food Nutrition 0.000 title claims abstract description 176
- 238000000034 method Methods 0.000 title claims abstract description 33
- 239000000463 material Substances 0.000 claims abstract description 32
- 238000013528 artificial neural network Methods 0.000 claims abstract description 26
- 238000013461 design Methods 0.000 claims abstract description 20
- 238000001914 filtration Methods 0.000 claims abstract description 17
- 238000003062 neural network model Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 38
- 230000008859 change Effects 0.000 claims description 15
- 238000001514 detection method Methods 0.000 claims description 12
- 235000015219 food category Nutrition 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 230000001902 propagating effect Effects 0.000 claims description 6
- 238000012790 confirmation Methods 0.000 claims description 5
- 238000009499 grossing Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 claims description 3
- 239000004744 fabric Substances 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 claims description 3
- 230000006855 networking Effects 0.000 claims description 3
- 238000004171 remote diagnosis Methods 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 abstract description 10
- 235000016709 nutrition Nutrition 0.000 description 7
- 230000035764 nutrition Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 241000282326 Felis catus Species 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 235000001014 amino acid Nutrition 0.000 description 1
- 150000001413 amino acids Chemical class 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 235000019784 crude fat Nutrition 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000019621 digestibility Nutrition 0.000 description 1
- 230000006806 disease prevention Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000005176 gastrointestinal motility Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 235000013402 health food Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 235000021056 liquid food Nutrition 0.000 description 1
- 244000144972 livestock Species 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 235000013372 meat Nutrition 0.000 description 1
- 235000010755 mineral Nutrition 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 230000000050 nutritive effect Effects 0.000 description 1
- 235000021395 porridge Nutrition 0.000 description 1
- 244000144977 poultry Species 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 235000011888 snacks Nutrition 0.000 description 1
- 230000004936 stimulating effect Effects 0.000 description 1
- 239000011573 trace mineral Substances 0.000 description 1
- 235000013619 trace mineral Nutrition 0.000 description 1
- 235000013343 vitamin Nutrition 0.000 description 1
- 239000011782 vitamin Substances 0.000 description 1
- 229940088594 vitamin Drugs 0.000 description 1
- 229930003231 vitamin Natural products 0.000 description 1
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
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P60/00—Technologies relating to agriculture, livestock or agroalimentary industries
- Y02P60/80—Food processing, e.g. use of renewable energies or variable speed drives in handling, conveying or stacking
- Y02P60/87—Re-use of by-products of food processing for fodder production
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method and a system for customizing pet food based on personalized recommendation, wherein the method comprises the following steps: s1, receiving a pet food searching instruction of a user needing to search; s2, fusing a pet food searching instruction of a user and attributes of pet food by using an Autoencoder neural network model; s3, clustering pet food searching instructions of the user by using a k-means algorithm; s4, performing collaborative filtering on each cluster obtained by clustering, and generating recommendations to required users; s5, providing picture materials by the needed products selected by the user; s6, the designer realizes effective identification of the image materials through Python language programming; and S7, generating a customized recommended pet food design scheme. The method uses an Autoencoder dimensionality reduction clustering collaborative filtering algorithm, replaces the required user-pet food store scoring vector with low-dimensional neural network characteristics with stronger expression capability, and improves the recommendation result.
Description
Technical Field
The invention relates to a method and a system for customizing pet food based on personalized recommendation.
Background
The pet food is a food specially provided for pets and small animals, is a high-grade animal food between human food and traditional livestock and poultry feed, and mainly has the function of providing basic nutrient substances required for life guarantee, growth and health of various pets. Has the advantages of comprehensive nutrition, high digestibility, scientific formula, standard quality, convenient feeding and use, certain disease prevention and the like.
According to the form division of pet food, the pet food is divided into dry type pet food, such as: fish food, dog food, cat food, snack food; semi-dry pet foods, such as: canned pet food, homemade dog food and cat food; pet liquid foods, such as: pet meat paste, pet nutritional porridge, and the like; according to the use of pet foods, pet foods are divided into: pet ration, a ration companion, pet health food, pet treats, a formula food, and the like.
The standard of pet food covers the contents of moisture, protein, crude fat, crude ash, crude fiber, nitrogen-free extract, mineral substances, trace elements, amino acids, vitamins and the like, wherein the crude ash is non-nutritive content, and the crude fiber has the function of stimulating gastrointestinal motility.
The nutrition design and manufacture of the pet food are guided by pet nutriologists of pet nutriology professionals, and scientific and reasonable pet food standards are made according to nutrition requirements by comprehensively considering the contents of the pets in different growth stages, self constitutions, different seasons and the like. When the food is selected and used for pets, the food is selected according to the physiological characteristics and growth stages of the pets, and reasonably matched and fed.
At present, by 6.1.2018, national quality standards about pet foods are not issued in China, the pet foods generally refer to related Chinese feed standards and American AAFCO pet nutrition standards, individual indexes refer to human food standards, and the pet foods enter families owning pets in cities at the fastest speed along with the high-speed development of Chinese economy and the update of modern consumption concepts.
With the high-speed development of the internet technology, people break through the space-time limitation of the traditional life style, and contact more and more information, and the ways of contacting the information are more and more abundant. Such as blogs, micro blogs, web portals, public numbers, news clients, encyclopedias, and e-commerce web sites. Although the massive Information on the internet brings unprecedented convenience and opportunity for users, the massive Information on the internet also brings unprecedented challenges for users and Information retrieval technologies, the challenges are mainly 'Information overload', the so-called 'Information overload' mainly means that the users are difficult to rapidly obtain valuable Information from a large amount of data, in the prior art, article recommendation is mainly performed by using a collaborative filtering method, but when the user has few mark samples, the precision of prediction is not high only by using the mark samples, and effective recommendation cannot be performed.
When purchasing pet food, a user may have a large number of suppliers and products for the pet food to choose from, and a great deal of time and effort may be expended in selecting pet food to select the product that best suits the needs of a particular pet. However, veterinarians and other specialists may assist in selecting a particular brand of pet food for a particular pet, meeting average pet needs in a selectable range of pets, the selection range being divided according to the age and or individual of the pet. However, the nutritional requirements for different pets vary, and a personalized system tailored to the nutrition of a particular pet or to the feeding of the pet would be advantageous. However, due to the practical difficulty and expense of matching individual foods for specific pets, users are often forced to choose from a limited number of different types of mass-produced pet foods and products, and merchants cannot effectively recommend their specific products to specific users, so that both buyers and sellers have trouble in shopping.
An effective solution to the problems in the related art has not been proposed yet.
Disclosure of Invention
Aiming at the problems in the related art, the invention provides a customized pet food method and a customized pet food system based on personalized recommendation, so as to overcome the technical problems in the prior related art.
Therefore, the invention adopts the following specific technical scheme:
according to one aspect of the invention, a customized pet food method based on personalized recommendation is provided, and the customized pet food method based on personalized recommendation comprises the following steps:
s1, receiving a pet food searching instruction of a user needing to search for pet food, wherein the instruction comprises a pet food category;
s2, fusing the pet food searching instruction of the user and the attributes of the pet food by using an Autoencoder neural network model, smoothing the vector between the pet food searching instruction of the user and the scores of the pet food, and training out low-dimensional and dense features;
s3, clustering the pet food search instructions of the user by using the characteristics of a k-means algorithm based on an Autoencoder neural network;
s4, carrying out collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering, and generating recommendation to the required user;
s5, selecting a proper required product according to the recommendation by the user, providing a picture material for the selected required product, transmitting the picture material to customer service personnel on line, and transmitting the picture material to a designer after the customer service personnel finishes the picture material;
s6, a designer realizes effective identification aiming at image materials through Python language programming, intelligently adds alternative pet food icons of a database and scales the alternative pet food icons to a proper size to generate point location design;
s7, the computer synchronously outputs the marked required pet food quantity, brand and model to generate a formatted list and generates a personalized recommended customized pet food design scheme.
Further, the fusing the pet food search instruction of the user and the attributes of the pet food by using the Autoencoder neural network model, smoothing a vector between the pet food search instruction of the user and the score of the pet food, and training low-dimensional and dense features further comprises the following steps:
s21, giving unlabelled data of a pet food search instruction of a user in need, and performing unsupervised learning data characteristics;
s22, inputting the unlabeled data into an encoder to obtain a coded code, decoding the code by a decoder to output data information, and calculating the reconstruction error of the data information and the unlabeled data;
s23, adjusting parameters of an encoder and a decoder to minimize the reconstruction error, wherein the encoding code is the representation of the input signal;
s24, using the coding code generated by the encoder as the next layer input, repeating the training of the steps S21-S23, and sequentially training layer by layer to obtain a multilayer network structure;
and S25, training a network by adopting the labeled data to the multilayer network structure through the reverse neural network, training out low-dimensional and dense features, and obtaining a new training sample set.
Further, the training of the network by the inverse neural network using the labeled data for the multi-layer network structure and training of the low-dimensional and dense features further includes the following steps:
s251, initializing connection weight and a threshold value by random small values;
s253, forward propagating the output of the computing node;
s254, calculating an error between the expected output and the actual output of the network structure;
s255, reversely propagating and adjusting each connection weight of the network structure;
s255, taking another group of samples, repeating the steps S252 to S255 until the error of the input and output samples reaches the requirement, and obtaining a new training sample set。
Further, the clustering the pet food search instruction of the user by using the characteristics obtained by the k-means algorithm based on the Autoencoder neural network further comprises the following steps:
s31, according to the new training sample setPartitioning of clustered clusters using k-means algorithm;
And S32, optimizing and minimizing the square error of the cluster by utilizing the optimization target.
Further, the square error formula is as follows:
wherein E is the square error;
the smaller the value, the higher the similarity of the samples within the cluster,the larger the value, the lower the similarity of the samples within the cluster.
Further, the optimizing the square error of the minimized cluster by using the optimization objective further includes the following steps:
s321, inputting parameters m and k, and selecting k centroids by using a random calculation method;
s322, calculating the distances from all data points to the cluster centroid;
s323, distributing the data points to the centroids with the nearest distance;
s324, updating the cluster centroid according to the change of the centroid;
and S325, when the centroid of the cluster centroid is not changed any more, the cluster is the optimized minimized cluster.
Further, the collaborative filtering based on the pet food search instruction of the user and generating the recommendation to the required user for each cluster obtained by clustering further comprises the following steps:
s41, searching K neighbors of the user through the Pearson similarity, and constructing a K neighbor set;
S43, calculating a set of pet groceries for which the user is required to not scoreThe prediction score of each term of (a);
and S44, putting the pet food search instruction recommendation of the user with the highest predictive score into a recommendation set, and generating a recommendation to the required user.
Further, the calculation formula of the prediction score is as follows:
in the formula, u is a user;
uk is an adjacent user;
Further, the computer synchronously outputs the formatted list generated by the quantity, the brand and the model of the marked required pet food, and generates the customized recommended pet food design scheme, which further comprises the following steps:
s71, establishing a whole-process tracing mechanism from pet food icon type selection, order confirmation, transaction completion, equipment installation completion to user terminal online;
s72, according to the unique identifier of the equipment networking, establishing an online and offline unified database, and completing the unique identifier authentication;
s73, detecting the state of each device, namely active detection and passive detection, wherein the active detection is a remote time-sharing segmented reading state, and the passive detection is state change and automatically uploads the state change to system updating;
s74, establishing a serial number cluster of fault codes which are not connected with equipment, unstable in connection and partial in functional failure, and updating, uploading and downloading in real time;
s75, remotely resetting and solving the fault problem which belongs to the resetsolvable problem;
s76, a device tracing mechanism compares whether the device in the database is over-guaranteed to solve the problem of whether the device is charged for internal and external protection;
s77, according to the fault code, remote diagnosis, online and offline consultation can be performed;
and S78, realizing one-time maintenance or replacement service according to the diagnosis result.
According to another aspect of the present invention, there is also provided a customized pet food system based on personalized recommendations, the system comprising:
the receiving instruction module is used for receiving a pet food searching instruction of a user needing the pet food searching instruction, and the instruction comprises a pet food category;
the Autoencorder neural network model fusion module fuses the pet food search instruction of the user and the attributes of the pet food by using the Autoencorder neural network model, smoothes the vector between the pet food search instruction of the user and the score of the pet food, and trains low-dimensional and dense features;
the k-means algorithm clustering module is used for clustering the pet food searching instructions of the user by utilizing the characteristics of the k-means algorithm based on the Autoencoder neural network;
the collaborative filtering recommendation module is used for performing collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering and generating recommendation to the required user;
the system comprises a user selection module, a designer and a user selection module, wherein the user selects a proper required product according to recommendation, provides a picture material for the selected required product, transmits the picture material to the customer service staff on line, and transmits the picture material to the designer after the customer service staff sorts the image material;
the pet food customizing module is used for enabling designers to realize effective identification aiming at image materials through Python language programming, intelligently adding alternative pet food icons of the database and zooming to a proper size to generate point location design;
and the computer synchronously outputs a formatted list generated by the quantity, the brand and the model of the marked required pet food, and generates a customized pet food design scheme recommended by the individuation.
The invention has the beneficial effects that:
1. the method utilizes the clustering collaborative filtering algorithm of Autoencoder dimensionality reduction, replaces the previous required user-pet food store scoring vector with the low-dimensional neural network characteristic with stronger expression capability, or replaces the required user-pet food store scoring vector after dimensionality reduction, and thus, the data dimensionality reduction and characteristic enhancement effects can be achieved; clustering by using the characteristics of the Autoencorder neural network, wherein the clustering can obtain a better clustering result, thereby achieving the effect of improving the algorithm precision; after the Autoencorder neural network obtains the low-dimensional features, a trained feature dimension reduction model is also obtained, so that the model does not need to be retrained under the condition that a pet food store database does not change greatly, other dimension reduction methods need to retrain the model when data is updated, and the model is more flexible and practical; the combination of the Autoencorder neural network and the clustering can further improve the expansibility of the model, effectively capture the characteristic of the required user preference change, apply the dynamic change of the preference to personalized recommendation and improve the effectiveness of the recommendation result.
2. The method comprises the steps of achieving effective identification aiming at a plane drawing through Python language programming, intelligently adding an alternative product icon of a database and zooming to a proper size, generating point location design, and synchronously outputting a formatted list generated by the number, the brand and the model of marked equipment by a system; the input requirements set by the user at the client are obtained, and the personalized pet food customization scheme of the user is generated according to the confirmation of the type selection of the user at the client, so that the personalized customization requirements of the user are met.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described 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 without creative efforts.
FIG. 1 is a flow chart of a method for customizing pet food based on personalized recommendations, in accordance with an embodiment of the present invention;
FIG. 2 is a diagram of a network model architecture of an Autoencor in a customized pet food method based on personalized recommendations, according to an embodiment of the present invention.
Detailed Description
For further explanation of the various embodiments, the drawings which form a part of the disclosure and which are incorporated in and constitute a part of this specification, illustrate embodiments and, together with the description, serve to explain the principles of operation of the embodiments, and to enable one skilled in the art to understand the embodiments and advantages of the disclosure for reference and without scale, wherein elements are not shown in the drawings and like reference numerals are used to refer to like elements generally.
According to the embodiment of the invention, a customized pet food method and system based on personalized recommendation are provided.
Referring now to the drawings and the detailed description, in accordance with an embodiment of the present invention, a customized pet food method based on personalized recommendation, as shown in fig. 1, comprises the following steps:
s1, receiving a pet food searching instruction of a user needing to search for pet food, wherein the instruction comprises a pet food category;
s2, fusing the pet food searching instruction of the user and the attributes of the pet food by using an Autoencoder neural network model, smoothing the vector between the pet food searching instruction of the user and the scores of the pet food, and training out low-dimensional and dense features;
in one embodiment, wherein the whole step S2 is an application of one of the Autoencoder neural network models, the fusing the pet food search command of the user and the attribute of the pet food and smoothing the vector between the pet food search command of the user and the score of the pet food by using the Autoencoder neural network model, and training the low-dimensional and dense features further comprises the following steps:
s21, giving unlabelled data of a pet food search instruction of a user in need, and performing unsupervised learning according to the unlabelled data to obtain data characteristics in the unlabelled data;
s22, inputting the unlabeled data into an encoder to obtain a coded code, decoding the code by a decoder to output data information, and calculating a reconstruction error between the data information and the unlabeled data;
s23, adjusting parameters of an encoder and a decoder to minimize the reconstruction error, wherein the encoding code is the representation of the input signal;
wherein, the parameter adjustment of the encoder and the decoder outputs data information according to the code decoding of the decoder, and the parameter adjustment of the encoder and the decoder is carried out artificially according to the result obtained in the step S22;
s24, using the coding code generated by the encoder as the next layer input, repeating the training of the steps S21-S23, and sequentially training layer by layer to obtain a multilayer network structure;
s25, training a network by adopting labeled data to a multilayer network structure through a reverse neural network, training out low-dimensional and dense features, and obtaining a new training sample set;
wherein, the whole steps S22-S25 are the concrete steps of unsupervised learning and data feature acquisition in S21;
in one embodiment, the training the network by using the labeled data for the multi-layer network structure through the inverse neural network, and training out the low-dimensional and dense features further comprises the following steps:
s251, initializing connection weight and a threshold value by random small values;
s253, forward propagating the output of the computing node;
s254, calculating an error between the expected output and the actual output of the network structure;
s255, reversely propagating and adjusting each connection weight of the network structure;
s255, taking another group of samples, repeating the steps S252 to S255 until the error of the input and output samples reaches the requirement, and obtaining a new training sample set;
The steps S251-S255 use a BP algorithm (the BP algorithm consists of two processes of signal forward propagation and error backward propagation), and the BP algorithm is used for training a network by adopting labeled data to a multi-layer network structure through a backward neural network, training low-dimensional and dense features and obtaining a new training sample set;
in a specific application, an automatic encoder (Autoencoder) is an unsupervised feature learning algorithm based on an error inverse propagation algorithm, and the algorithm mainly utilizes a back propagation algorithm to make a target value equal to an input value as much as possible. Autoencoder is understood to be the process of encoding (encoder) -decoding (decoder), and we put the input features into an encoder of encoder to get an encoding (code), i.e. the representation of the input features in low dimension, and then add it to a decoder of decoder. We want the decoder to be able to output as much as possible the same features as the input, so that the resulting code is a good representation of the features, we minimize the reconstruction error by adjusting the parameters of the encoder and the decoder. The architecture of the Autoencoder network is shown in fig. 2, and belongs to a three-layer neural network, where the first layer is an input layer, the second layer is a hidden layer (also referred to as a feature layer), and the third layer is an output layer. Its learning goal is to learn an identity functionSo as to outputAs much as possible equal to the input. For example, given a high-dimensional feature with a dimension of 1000, the number of neurons in the hidden layer is 100, the Autoencoder needs to reconstruct 1000-dimensional input from the 100-dimensional hidden layer feature, and the 100-dimensional data of the hidden layer includes important features of the input data.
S3, clustering the pet food search instructions of the user by using the characteristics of a k-means algorithm based on an Autoencoder neural network;
in one embodiment, the clustering the pet food search instructions of the user based on the features obtained by the Autoencoder neural network using the k-means algorithm further comprises the following steps:
s31, according to the new training sample setPartitioning of clustered clusters using k-means algorithm;
S32, optimizing and minimizing the square error of the cluster by utilizing the optimization target;
in one embodiment, the square error formula is as follows:
wherein E is the square error;
the smaller the value, the higher the similarity of the samples within the cluster,the larger the value, the lower the similarity of the samples in the cluster;
in one embodiment, the optimizing the squared error of the minimized cluster using the optimization objective further comprises the steps of:
s321, inputting parameters m and k, and selecting k centroids by using a random calculation method, wherein m is a parameter and can be any numerical value;
s322, calculating the distances from all data points to the cluster centroid;
s323, distributing the data points to the centroids with the shortest distances;
s324, updating the cluster centroid according to the change of the centroid;
and S325, when the centroid of the cluster centroid is not changed any more, the cluster is the optimized minimized cluster.
Unsupervised learning (unsupervised learning) is the most important branch in the field of machine learning, and mainly learns the inherent connection and rule among unlabeled training samples, wherein clustering (clustering) is the most widely applied. The basic idea of clustering is to divide a data set into several mutually disjoint subsets, each subset being called a "cluster", and each data belonging to a "cluster". Common clustering algorithms are mainly classified into prototypical clustering (prototypical based clustering), which includes K-means (K-means) algorithm and Gaussian-of-Gaussian (Mixture-of-Gaussian) algorithm; density-based clustering includes DBSCAN; hierarchical clustering (hierarchical clustering) includes AGENS. The most widely used and successful clustering method is the K-means algorithm in the invention.
S4, carrying out collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering, and generating recommendation to the required user;
in one embodiment, the performing collaborative filtering based on pet food search instructions of the user for each cluster obtained by clustering and generating recommendations to the desired user further comprises the steps of:
s41, searching K neighbors of the user through the Pearson similarity, and constructing a K neighbor set;
S43, calculating a set of pet groceries for which the user is required to not scoreThe prediction score of each term of (a);
and S44, putting the pet food search instruction recommendation of the user with the highest predictive score into a recommendation set, and generating a recommendation to the required user.
In one embodiment, the predictive score is calculated as follows:
in the formula, u is a user;
uk is a neighboring user;
S5, selecting a proper required product by the user according to the recommendation, providing a picture material for the selected required product, transmitting the picture material to customer service personnel on line, and transmitting the picture material to a designer after the customer service personnel finishes the picture material;
s6, a designer realizes effective identification aiming at image materials through Python language programming, intelligently adds alternative pet food icons of a database and scales the alternative pet food icons to a proper size to generate point location design;
s7, the computer synchronously outputs the marked required pet food quantity, brand and model to generate a formatted list and generates a personalized recommended customized pet food design scheme.
In one embodiment, the computer synchronizing the output of the formatted list of tagged desired pet food quantities, brands, and models and generating the personalized recommended customized pet food design further comprises the steps of:
s71, establishing a whole-process tracing mechanism from pet food icon type selection, order confirmation, transaction completion, equipment installation completion to user terminal online;
s72, according to the unique identifier of the equipment networking, establishing an online and offline unified database, and completing the unique identifier authentication;
s73, detecting the state of each device, namely active detection and passive detection, wherein the active detection is a remote time-sharing segmented reading state, and the passive detection is state change and automatically uploads the state change to system updating;
s74, establishing a serial number cluster of fault codes which are not connected with equipment, unstable in connection and partial in functional failure, and updating, uploading and downloading in real time;
s75, remotely resetting and solving the fault problem which belongs to the resetsolvable problem;
s76, a device tracing mechanism compares whether the device in the database is over-guaranteed to solve the problem of whether the device is charged for internal and external protection;
s77, according to the fault code, remote diagnosis, online and offline consultation can be performed;
and S78, realizing one-time maintenance or replacement service according to the diagnosis result.
There is also provided, in accordance with another embodiment of the present invention, a customized pet food system based on personalized recommendations, the system including:
the receiving instruction module is used for receiving a pet food searching instruction of a user needing the pet food searching instruction, and the instruction comprises a pet food category;
the Autoencorder neural network model fusion module fuses the pet food search instruction of the user and the attributes of the pet food by using the Autoencorder neural network model, smoothes the vector between the pet food search instruction of the user and the score of the pet food, and trains low-dimensional and dense features;
the k-means algorithm clustering module is used for clustering the pet food searching instructions of the user by utilizing the characteristics of the k-means algorithm based on the Autoencoder neural network;
the collaborative filtering recommendation module is used for performing collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering and generating recommendation to the required user;
the system comprises a user selection module, a designer and a user selection module, wherein the user selects a proper required product according to recommendation, provides a picture material for the selected required product, transmits the picture material to the customer service staff on line, and transmits the picture material to the designer after the customer service staff sorts the image material;
the pet food customizing module is used for enabling designers to realize effective identification aiming at image materials through Python language programming, intelligently adding alternative pet food icons of the database and zooming the alternative pet food icons to a proper size to generate point location design;
the pet food design module is used for synchronously outputting a formatted list generated by the quantity, the brand and the model of the marked required pet food by the computer and generating an individualized and recommended customized pet food design scheme
In summary, with the aid of the above technical solution of the present invention, the present invention utilizes an automatic encoder dimension reduction clustering Collaborative Filtering Algorithm (AECCF), replaces the previous required user-pet food store scoring vector with a low-dimensional neural network feature with a stronger expression capability, or replaces the previous required user-pet food store scoring vector with the required user-pet food store scoring vector after dimension reduction, which can achieve the effect of reducing data dimension and enhancing feature; clustering by using the characteristics of the Autoencorder neural network, wherein the clustering can obtain a better clustering result, thereby achieving the effect of improving the algorithm precision; after the Autoencorder neural network obtains the low-dimensional features, a trained feature dimension reduction model is also obtained, so that the model does not need to be retrained under the condition that a pet food store database does not change greatly, other dimension reduction methods need to retrain the model when data is updated, and the model is more flexible and practical; the combination of the Autoencorder neural network and clustering can further improve the expansibility of the model, effectively capture the characteristic of the required user preference change, apply the dynamic change of the preference to personalized recommendation, improve the effectiveness of the recommendation result, realize effective identification aiming at the plane drawing through Python language programming, intelligently add the alternative product icons of the database and zoom to the proper size, generate point location design, and synchronously output a formatted list generated by the marked equipment number, brand and model by the system; the input requirements set by the user at the client are obtained, and the personalized pet food customization scheme of the user is generated according to the confirmation of the type selection of the user at the client, so that the personalized customization requirements of the user are met.
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 (9)
1. A customized pet food method based on personalized recommendation is characterized by comprising the following steps:
s1, receiving a pet food searching instruction of a user needing to search for pet food, wherein the instruction comprises a pet food category;
s2, fusing the pet food search instruction of the user and the attributes of the pet food by using an Autoencoder neural network model, smoothing the vector between the pet food search instruction of the user and the scores of the pet food, and training out low-dimensional and dense features:
s21, giving unlabelled data of a pet food search instruction of a user in need, and performing unsupervised learning data characteristics;
s22, inputting the non-label data into an encoder to obtain a code, decoding the code by a decoder to output data information, and calculating the reconstruction error of the data information and the non-label data;
s23, adjusting parameters of an encoder and a decoder to minimize the reconstruction error, wherein the encoding code is the representation of the input signal;
s24, using the coding code generated by the encoder as the next layer input, repeating the training of the steps S21-S23, and sequentially training layer by layer to obtain a multilayer network structure;
s25, training a network by adopting labeled data to a multilayer network structure through a reverse neural network, training out low-dimensional and dense features, and obtaining a new training sample set;
s3, clustering the pet food search instructions of the user by using the characteristics of a k-means algorithm based on an Autoencoder neural network;
s4, carrying out collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering, and generating recommendation to the required user;
s5, selecting a proper required product by the user according to the recommendation, providing a picture material for the selected required product, transmitting the picture material to customer service personnel on line, and transmitting the picture material to a designer after the customer service personnel finishes the picture material;
s6, a designer realizes effective recognition of image materials through Python language programming, intelligently adds alternative pet food icons of the database and zooms the alternative pet food icons to a proper size to generate a point location design;
s7, the computer synchronously outputs the marked required pet food quantity, brand and model to generate a formatted list and generates a personalized recommended customized pet food design scheme.
2. The method of claim 1, wherein the training of the multi-layered network structure with the labeled data through the inverse neural network to obtain the low-dimensional and dense features further comprises the steps of:
s251, initializing connection weight and a threshold value by random small values;
s253, forward propagating the output of the computing node;
s254, calculating an error between the expected output and the actual output of the network structure;
s255, reversely propagating and adjusting each connection weight of the network structure;
3. The method of claim 1, wherein clustering the pet food search instruction of the user based on the features obtained by the Autoencoder neural network using the k-means algorithm further comprises the steps of:
s31, according to the new training sample setPartitioning of clustered clusters using k-means algorithm;
And S32, optimizing and minimizing the square error of the cluster by utilizing the optimization target.
4. The customized pet food method based on personalized recommendations according to claim 3, wherein the square error formula is as follows:
wherein E is the square error;
5. The customized pet food method based on personalized recommendations according to claim 3, wherein said optimizing the squared error of the minimized cluster using optimization objectives further comprises the steps of:
s321, inputting parameters m and k, and selecting k centroids by using a random calculation method;
s322, calculating the distances from all the data points to the cluster centroid;
s323, distributing the data points to the centroids with the shortest distances;
s324, updating the cluster centroid according to the change of the centroid;
and S325, when the centroid of the cluster centroid is not changed any more, the cluster is the optimized minimized cluster.
6. The method of claim 1, wherein the step of performing a collaborative filtering based on pet food search instructions of the user for each cluster obtained by clustering and generating the recommendation to the desired user further comprises the steps of:
s41 finding out Pearson similarityFinding K neighbors of the user and constructing a K neighbor set;
S43, calculating the required pet food store set of user unscoredThe prediction score of each term of (a);
and S44, putting the pet food search instruction recommendation of the user with the highest predictive score into a recommendation set, and generating a recommendation to the required user.
7. The customized pet food method based on personalized recommendations according to claim 6, wherein the predictive score is calculated as follows:
in the formula, u is a user;
uk is an adjacent user;
8. The customized pet food method based on personalized recommendation of claim 1, wherein said computer synchronously outputs formatted lists generated by the number, brand and model of the marked required pet food, and generates customized recommended customized pet food design scheme further comprising the steps of:
s71, establishing a whole-process tracing mechanism from pet food icon type selection, order confirmation, transaction completion, equipment installation completion to user terminal online;
s72, according to the unique identifier of the equipment networking, establishing an online and offline unified database, and completing the unique identifier authentication;
s73, detecting the state of each device, namely active detection and passive detection, wherein the active detection is a remote time-sharing segmented reading state, and the passive detection is state change and automatically uploads the state change to system updating;
s74, establishing a serial number cluster of fault codes which are not connected with equipment, unstable in connection and partial in functional failure, and updating, uploading and downloading in real time;
s75, remotely resetting and solving the fault problem which belongs to the resetsolvable problem;
s76, a device tracing mechanism compares whether the device in the database is over-guaranteed to solve the problem of whether the device is charged for internal and external protection;
s77, according to the fault code, remote diagnosis, online and offline consultation can be performed;
and S78, realizing one-time maintenance or replacement service according to the diagnosis result.
9. A customized pet food system based on personalized recommendation, which is used for implementing the customized pet food method based on personalized recommendation of any one of claims 1-8, and is characterized in that the system comprises:
the receiving instruction module is used for receiving a pet food searching instruction of a user needing the pet food searching instruction, and the instruction comprises a pet food category;
the automatic encoder neural network model fusion module fuses the pet food search instruction of the user and the attributes of the pet food by using the automatic encoder neural network model, smoothes a vector between the pet food search instruction of the user and the score of the pet food, and trains out low-dimensional and dense features;
the k-means algorithm clustering module is used for clustering the pet food searching instructions of the user by utilizing the characteristics of the k-means algorithm based on the Autoencoder neural network;
the collaborative filtering recommendation module is used for performing collaborative filtering based on the pet food search instruction of the user on each cluster obtained by clustering and generating recommendation to the required user;
the system comprises a user selection module, a designer and a user selection module, wherein the user selects a proper required product according to recommendation, provides a picture material for the selected required product, transmits the picture material to the customer service staff on line, and transmits the picture material to the designer after the customer service staff sorts the image material;
the pet food customizing module is used for enabling designers to realize effective identification aiming at image materials through Python language programming, intelligently adding alternative pet food icons of the database and zooming to a proper size to generate point location design;
and the computer synchronously outputs a formatted list generated by the quantity, the brand and the model of the marked required pet food, and generates a customized pet food design scheme recommended by the individuation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210953287.4A CN115017419A (en) | 2022-08-10 | 2022-08-10 | Customized pet food method and system based on personalized recommendation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210953287.4A CN115017419A (en) | 2022-08-10 | 2022-08-10 | Customized pet food method and system based on personalized recommendation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115017419A true CN115017419A (en) | 2022-09-06 |
Family
ID=83065620
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210953287.4A Pending CN115017419A (en) | 2022-08-10 | 2022-08-10 | Customized pet food method and system based on personalized recommendation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115017419A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117726389A (en) * | 2023-09-09 | 2024-03-19 | 广东德澳智慧医疗科技有限公司 | Cross-border e-commerce product purchase guiding system based on clustering recommendation algorithm |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111562749A (en) * | 2020-04-27 | 2020-08-21 | 李建军 | AI-based general intelligent household scheme automatic design method and device |
CN112307332A (en) * | 2020-10-16 | 2021-02-02 | 四川大学 | Collaborative filtering recommendation method and system based on user portrait clustering and storage medium |
-
2022
- 2022-08-10 CN CN202210953287.4A patent/CN115017419A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111562749A (en) * | 2020-04-27 | 2020-08-21 | 李建军 | AI-based general intelligent household scheme automatic design method and device |
CN112307332A (en) * | 2020-10-16 | 2021-02-02 | 四川大学 | Collaborative filtering recommendation method and system based on user portrait clustering and storage medium |
Non-Patent Citations (1)
Title |
---|
CHARLOTTE77: "KMeans与深度学习自编码AutoEncoder结合提高聚类效果", 《HTTPS://WWW.CNBLOGS.COM/CHARLOTTE77/P/5366578》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117726389A (en) * | 2023-09-09 | 2024-03-19 | 广东德澳智慧医疗科技有限公司 | Cross-border e-commerce product purchase guiding system based on clustering recommendation algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gao et al. | Hierarchical attention network for visually-aware food recommendation | |
Vielzeuf et al. | Centralnet: a multilayer approach for multimodal fusion | |
CN111125530B (en) | Information flow recommendation method based on multi-type feature deep learning | |
CN112749339B (en) | Travel route recommendation method and system based on travel knowledge graph | |
Karpathy | Connecting images and natural language | |
CN110134868B (en) | Recommendation method based on user preference heterogeneity analysis | |
CN110866542B (en) | Depth representation learning method based on feature controllable fusion | |
CN111125422A (en) | Image classification method and device, electronic equipment and storage medium | |
Zhu et al. | Multi-modal deep analysis for multimedia | |
CN111680217A (en) | Content recommendation method, device, equipment and storage medium | |
WO2021139415A1 (en) | Data processing method and apparatus, computer readable storage medium, and electronic device | |
CN111434118B (en) | Apparatus and method for generating user interest information | |
Liu et al. | Learning socially embedded visual representation from scratch | |
CN112818251A (en) | Video recommendation method and device, electronic equipment and storage medium | |
CN111949885A (en) | Personalized recommendation method for scenic spots | |
CN115017419A (en) | Customized pet food method and system based on personalized recommendation | |
CN112948710A (en) | Graph neural network-based punishment education recommendation method, system and storage medium | |
Cinar et al. | Inferring user interests on social media from text and images | |
Bayoudh | A survey of multimodal hybrid deep learning for computer vision: Architectures, applications, trends, and challenges | |
Zhu et al. | Cross-modal retrieval: a systematic review of methods and future directions | |
CN114817692A (en) | Method, device and equipment for determining recommended object and computer storage medium | |
KR102546355B1 (en) | Method, apparatus and system for providing nutritional information based on fecal image analysis | |
Fosset et al. | Docent: A content-based recommendation system to discover contemporary art | |
CN115409579A (en) | Meta-learning-based responsive recommendation method, system and equipment | |
Lin et al. | Automatic image annotation via combining low-level colour feature with features learned from convolutional neural networks |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220906 |
|
RJ01 | Rejection of invention patent application after publication |