Disclosure of Invention
The embodiment of the invention provides a data generation method, a device, a medium and electronic equipment, which are used for providing high-value commodity data for a user and providing reliable references for shopping decisions of the user.
In a first aspect, an embodiment of the present invention provides a data generating method, including:
acquiring historical evaluation data of an object to be evaluated, and extracting evaluation keywords of the historical evaluation data, wherein the evaluation keywords are used for representing physical characteristics of the object to be evaluated;
Generating evaluation data corresponding to the evaluation keywords according to the evaluation keywords and a preset neural network model;
And displaying the evaluation data at a first preset position in an evaluation area of the object to be evaluated.
In one possible design, after displaying the evaluation data at a preset position in the evaluation area of the object to be evaluated, the method further includes:
acquiring a first evaluation instruction input for first evaluation data, wherein the evaluation data comprises the first evaluation data;
And displaying a first sub-evaluation corresponding to the first evaluation instruction at a second preset position in an evaluation area of the object to be evaluated, wherein the first sub-evaluation is used for evaluating the first evaluation data, and the second preset position and the first preset position are adjacently arranged in the evaluation area.
In one possible design, the extracting the evaluation keywords of the historical evaluation data includes:
Performing word segmentation processing on the historical evaluation data according to a preset word stock to obtain a word segmentation set;
Calculating a weight value of each word in the word segmentation set according to a preset keyword weight algorithm;
selecting the segmentation word with the weight value larger than a preset weight value as a first evaluation keyword of the historical evaluation data, wherein the evaluation keyword comprises the first evaluation keyword.
In one possible design, the obtaining historical evaluation data of the object to be evaluated includes:
when the number of the items of the evaluation corresponding to the object to be evaluated is larger than a preset threshold value of the number of the items, acquiring the current evaluation as the historical evaluation data corresponding to the object to be evaluated.
In one possible design, before the word segmentation processing is performed on the historical evaluation data according to a preset word stock, the method further includes:
preprocessing the historical evaluation data, wherein the preprocessing comprises removing stop words and punctuation marks in the historical evaluation data.
In one possible design, the obtaining the historical evaluation data of the object to be evaluated, and extracting the evaluation keywords of the historical evaluation data, includes:
and when the evaluation item corresponding to the object to be evaluated is smaller than or equal to the preset item threshold value, acquiring an evaluation keyword corresponding to the three-level category corresponding to the object to be evaluated as a second evaluation keyword, wherein the evaluation keyword comprises the second evaluation keyword.
In one possible design, the generating, according to the evaluation keyword and a preset neural network model, evaluation data corresponding to the evaluation keyword includes:
generating first evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
generating second evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
Wherein the evaluation data includes the first evaluation data and the second evaluation data.
In one possible design, before the generating the evaluation data corresponding to the evaluation keyword according to the evaluation keyword and a preset neural network model, the method further includes:
Training the preset neural network model according to the evaluation keywords and the historical evaluation data, wherein the preset neural network model is a long-term and short-term memory network model.
In one possible design, the training the preset neural network model according to the evaluation keyword includes:
Vectorizing the evaluation keywords and each word in the word segmentation set, and inputting the vectorized data matrix as training data to train the long-term and short-term memory network model.
In a second aspect, an embodiment of the present invention further provides a data generating apparatus, including:
The acquisition module is used for acquiring historical evaluation data of the object to be evaluated;
the extraction module is used for extracting evaluation keywords of the historical evaluation data, wherein the evaluation keywords are used for representing the physical characteristics of the object to be evaluated;
The processing module is used for generating evaluation data corresponding to the evaluation keywords according to the evaluation keywords and a preset neural network model;
and the display module is used for displaying the evaluation data at a first preset position in the evaluation area of the object to be evaluated.
In one possible design, the obtaining module is further configured to obtain a first evaluation instruction input for first evaluation data, where the evaluation data includes the first evaluation data;
The display module is further configured to display a first sub-evaluation corresponding to the first evaluation instruction at a second preset position in the evaluation area of the object to be evaluated, where the first sub-evaluation is used to evaluate the first evaluation data, and the second preset position and the first preset position are adjacently arranged in the evaluation area.
In one possible design, the extraction module is specifically configured to:
Performing word segmentation processing on the historical evaluation data according to a preset word stock to obtain a word segmentation set;
Calculating a weight value of each word in the word segmentation set according to a preset keyword weight algorithm;
selecting the segmentation word with the weight value larger than a preset weight value as a first evaluation keyword of the historical evaluation data, wherein the evaluation keyword comprises the first evaluation keyword.
In one possible design, the acquisition module is specifically configured to:
when the number of the items of the evaluation corresponding to the object to be evaluated is larger than a preset threshold value of the number of the items, acquiring the current evaluation as the historical evaluation data corresponding to the object to be evaluated.
In one possible design, the processing module is further configured to pre-process the historical evaluation data, where the pre-process includes removing stop words and punctuation marks from the historical evaluation data.
In one possible design, the extraction module is specifically configured to:
and when the evaluation item corresponding to the object to be evaluated is smaller than or equal to the preset item threshold value, acquiring an evaluation keyword corresponding to the three-level category corresponding to the object to be evaluated as a second evaluation keyword, wherein the evaluation keyword comprises the second evaluation keyword.
In one possible design, the processing module is specifically configured to:
generating first evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
generating second evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
Wherein the evaluation data includes the first evaluation data and the second evaluation data.
In one possible design, the data generating device further includes:
and the training module is used for training the preset neural network model according to the evaluation keywords and the historical evaluation data, wherein the preset neural network model is a long-term and short-term memory network model.
In one possible design, the training module is specifically configured to:
Vectorizing the evaluation keywords and each word in the word segmentation set, and inputting the vectorized data matrix as training data to train the long-term and short-term memory network model.
In a third aspect, an embodiment of the present invention further provides an electronic device, including:
Processor, and
A memory for storing executable instructions of the processor;
Wherein the processor is configured to perform any one of the possible data generation methods of the first aspect via execution of the executable instructions.
In a fourth aspect, embodiments of the present invention also provide a storage medium having stored thereon a computer program which, when executed by a processor, implements any one of the possible data generation methods of the first aspect.
According to the data generation method, the device, the medium and the electronic equipment, the historical evaluation data of the object to be evaluated are obtained, the physical characteristic evaluation keywords which can be used for representing the object to be evaluated are extracted from the historical evaluation data, and then the evaluation data are obtained by inputting the evaluation keywords into the preset neural network model, wherein the evaluation data can be used as core evaluation data of the object to be evaluated and displayed at a first preset position in an evaluation area of the object to be evaluated, so that high-value evaluation data used for representing relevant characteristics of the object to be evaluated are output for a user.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of 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 apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Generally, the process of online purchasing an article by a user can be simplified into the following processes of (1) unintentionally browsing the article or purposefully searching for an article, (2) checking the detail of the article, (3) checking the comment of the article, and (4) ordering payment for purchase. In the step (2), assuming that the user is satisfied with the characteristics (shape style, material texture, price) of the merchandise, the user does not directly enter the purchasing link as in the case of shopping in a physical store, because the user understands that the merchandise seen at this time may be different from the real merchandise, the user will typically carefully browse comments about the merchandise by the purchased user, and pay attention to some characteristics such as color, style, etc. that are of interest, in this process, the merchandise comment information determines the ordering purchasing behavior of the user to a great extent.
At present, commodity comment information of online purchased commodities is voluntarily filled in by a user after the transaction is completed. Through simple analysis of the comment behaviors of the users, commodity evaluation on the Jingdong mall APP can be used as a reference, and in order to encourage the users to provide comment information, the Jingdong mall uses an incentive strategy that evaluation orders send Beijing beans.
Specifically, in this scenario, users can be roughly classified into three categories, the first category being non-participating reviewers, the second category being real reviewers, and the third category being the natto rewarders.
It can be understood that the first class of people has some people without evaluation consciousness, some people have evaluation consciousness, but feel that writing comments is a time-consuming and labor-consuming matter, while the second class of people mainly comes from people with no rewards and evaluation, the second class of people has stronger autonomous evaluation consciousness, generally has clear likes and dislikes to purchased commodities and wants to share the comments to the subsequent buyers, therefore, the comment information value submitted by the people is higher, and the third class of people is mainly people with random comments for rewards and general nonsensical comments.
In addition, the content of commodity comment information filled in by the user is different from person to person, and some comments are concise and emphasized, and some questions are not worth noting. Therefore, excessive commodity comment information can cause trouble to users when making purchase decisions, and effective evaluation data cannot be obtained quickly.
In view of the above-mentioned problems, an embodiment of the present invention provides a data generating method, by acquiring historical evaluation data of an object to be evaluated, extracting a physical characteristic evaluation keyword capable of being used for characterizing the object to be evaluated from the historical evaluation data, and then acquiring commodity evaluation data by inputting the evaluation keyword into a preset neural network model, wherein the commodity evaluation data can be used as core evaluation data of the object to be evaluated and displayed at a first preset position in an evaluation area of the object to be evaluated, thereby providing high-value commodity evaluation data for a user and providing reliable reference for shopping decisions of the user. Alternatively, the object to be evaluated may be understood as an article to be evaluated, and the history evaluation data may be understood as history evaluation information of the article.
The device realized by the data generation method in the scheme can be any electronic equipment capable of playing games, including tablet computers, smart phones, personal computers and the like. The data generation method is described in detail below in terms of several specific implementations.
Fig. 1 is a schematic view of an application scenario architecture of a data generation method according to an exemplary embodiment of the present invention. As shown in fig. 1, when it is required to provide evaluation data for an object to be evaluated, for example, when it is required to provide commodity evaluation data for a commodity to be evaluated, a request may be sent through a gateway, and then a corresponding core evaluation may be acquired from a core evaluation database through a commodity evaluation acquisition service, and these core evaluations may be used as commodity evaluation data of the commodity to be evaluated, and in response output, may be displayed in an evaluation area of the commodity to be evaluated.
The commodity evaluation obtaining service may be obtained from a core comment database through a commodity evaluation obtaining interface, or may be obtained from a history comment obtaining interface according to the history comment information of the current commodity to be evaluated.
It should be understood that, for obtaining from the historical comment obtaining interface according to the current historical comment information of the commodity to be evaluated, when the commodity to be evaluated already has a certain amount of historical evaluation data, for example, an evaluation number threshold may be set, when the evaluation number of the commodity exceeds the threshold, the commodity to be evaluated may be subjected to evaluation keyword extraction, that is, all the evaluation data are summarized into multiple evaluation sets, for example, 100 evaluation sets in each evaluation set, and then input into a trained Long Short-Term Memory network model (LSTM) to generate evaluation data, and a mapping relationship may be established between the evaluation data and the code of the commodity to be evaluated for subsequent query.
It is also worth understanding, and some commodities are considered to be just put on shelf, no historical comment information or too little comment information is available, and at the moment, comment keywords with three-level purposes are used as comment keywords of the commodities to be evaluated. In this case, the core comment database may be generated by a commodity evaluation generating service, and the commodity evaluation generating service may be online or offline, specifically, by first acquiring commodity information and historical evaluation data of the class of the corresponding commodity to be evaluated, then acquiring evaluation keywords in the historical evaluation data by a keyword weight algorithm, for example, a Term-inverse text Frequency index algorithm (Term-Inverse Document Frequency, abbreviated as TF-IDF), to generate an evaluation keyword database, then inputting the evaluation keywords into the LSTM to generate evaluation data, and further, may establish a mapping relationship between the commodity evaluation data and the class of the third, so as to facilitate subsequent query.
In addition, the commodity evaluation data corresponding to the evaluation keywords may be generated by inputting the evaluation keywords obtained in the two ways into the LSTM. The commodity evaluation data is coded with (object to be evaluated: core evaluation 1 the number of the samples was calculated, core evaluation 2......) and (tertiary categories: core rating 1, core rating 2.) is saved in a database for use in online queries.
Fig. 2 is a flow chart illustrating a data generation method according to an example embodiment of the present invention. As shown in fig. 2, the data generating method provided in this embodiment includes:
step 101, acquiring historical evaluation data of an object to be evaluated.
Specifically, when the number of items of the evaluation corresponding to the object to be evaluated is greater than a preset threshold value of the number of items, the current commodity evaluation is obtained as the historical evaluation data corresponding to the object to be evaluated. For example, when the number of items of evaluation data corresponding to the object to be evaluated is greater than 100, the commodity evaluation data may be used as historical evaluation data for extraction of evaluation keywords. The extraction may also be performed by combining all the historical evaluation data into a plurality of evaluation data sets, for example, each evaluation data set may include 100 comments.
And 102, extracting evaluation keywords of the historical evaluation data.
When the number of the items of the evaluation corresponding to the commodity to be evaluated is larger than a preset threshold value of the number of the items, after the historical evaluation data of the object to be evaluated is obtained, the evaluation keywords of the historical evaluation data can be continuously extracted, wherein the evaluation keywords are used for representing the physical characteristics of the commodity to be evaluated.
Specifically, word segmentation processing may be performed on the historical evaluation data according to a preset word bank to obtain a word segmentation set, then, a weight value of each word segment in the word segmentation set is calculated according to a preset keyword weight algorithm, and the word segment with the weight value larger than the preset weight value is selected as a first evaluation keyword of the historical evaluation data, wherein the evaluation keyword comprises the first evaluation keyword. Optionally, in order to divide words more accurately, before the historical evaluation data is divided according to the preset word stock, the historical evaluation data can be preprocessed, wherein the preprocessing can include removing stop words and punctuation marks in the historical evaluation data.
It should be noted that, in this embodiment, the preset word stock may be a bargain word stock or other word stocks, or may be an extended word stock added with a custom dictionary based on the existing word stock. And, the weighting algorithm for the keywords may be a TF-IDF algorithm. The principle of the TF-IDF algorithm is that the product of TF and IDF is calculated, and then the importance degree of words in a word stock on each document is measured by using the result of the product to evaluate the importance degree of keywords on one document in a document set or a corpus. The specific implementation principle of the TF-IDF algorithm is an existing algorithm, and is not specifically limited in this embodiment.
In addition, in another possible case, when some commodities are just put on shelf and there is no history evaluation data or the evaluation data is too small, that is, when the evaluation item corresponding to the commodity to be evaluated is less than or equal to the preset item threshold value, the evaluation keyword corresponding to the three-level category corresponding to the commodity to be evaluated may be acquired as the second evaluation keyword, and the evaluation keyword includes the second evaluation keyword.
And step 103, generating evaluation data corresponding to the evaluation keywords according to the evaluation keywords and a preset neural network model.
And after the evaluation keywords are obtained, the evaluation keywords can be input into the LSTM model after training is completed so as to generate evaluation data corresponding to the evaluation keywords. The training of the preset neural network model, such as the LSTM model, is described in detail in the following embodiments. The model can be trained through a deep learning mode, the deep learning is a machine learning algorithm, and the deep learning can be understood as a data characteristic learning method. The method uses multi-layer nonlinear processing unit cascade to extract and convert the characteristics, and each continuous layer uses the output of the previous layer as input, so that the higher-level data characteristics can be learned by abstracting the characteristics layer by layer. And LSTM is a special cyclic neural network, and the LSTM solves the problem that the traditional cyclic neural network is difficult to process long-distance dependence by adding an input threshold, a forgetting threshold and an output threshold.
And 104, displaying commodity evaluation data at a first preset position in an evaluation area of the object to be evaluated.
After generating the evaluation data corresponding to the evaluation keywords according to the evaluation keywords and the preset neural network model, the commodity evaluation data can be displayed at a first preset position in the evaluation area of the commodity to be evaluated so as to evaluate the commodity to be evaluated.
In one possible scenario, FIG. 3 is a schematic diagram of one possible merchandise information browsing interface in the embodiment shown in FIG. 2. The mobile phone may be selected as an example of the commodity to be evaluated, as shown in fig. 3, the first preset position in the evaluation area may be the core evaluation area shown in fig. 3, and the commodity evaluation data generated by the method of the embodiment may be, for example, complete functions, comfortable use, perfect operation, etc. The commodity evaluation data in the core evaluation area may be core evaluation data determined according to the historical evaluation data in all the evaluation areas, or core evaluation data corresponding to the commodity class III (mobile phone).
In this embodiment, the historical evaluation data of the commodity to be evaluated is obtained, the physical characteristic evaluation keywords capable of being used for representing the commodity to be evaluated are extracted from the historical evaluation data, and then the commodity evaluation data is obtained by inputting the evaluation keywords into a preset neural network model, wherein the commodity evaluation data can be used as core evaluation data of the commodity to be evaluated and displayed at a first preset position in an evaluation area of the commodity to be evaluated, so that high-value evaluation data for representing relevant characteristics of an object to be evaluated is output for a user.
Fig. 4 is a flow chart illustrating a data generation method according to another example embodiment of the present invention. As shown in fig. 4, the data generating method provided in this embodiment includes:
step 201, acquiring historical evaluation data of an object to be evaluated.
And 202, extracting evaluation keywords of the historical evaluation data.
And 203, generating evaluation data corresponding to the evaluation keywords according to the evaluation keywords and a preset neural network model.
And 204, displaying commodity evaluation data at a first preset position in an evaluation page of the object to be evaluated.
It should be noted that, the specific implementation manner of the steps 201 to 204 in this embodiment refers to the descriptions of the steps 101 to 104 in the embodiment shown in fig. 2, and will not be repeated here.
Step 205, a first evaluation instruction for the first evaluation data input is acquired.
After the commodity evaluation data are displayed at the first preset position in the evaluation page of the commodity to be evaluated, the commodity evaluation is more conveniently carried out by the user, and particularly, the frequency of carrying out the evaluation price of the first class user is improved, and the credibility of the evaluation data of the third class user is improved. The first evaluation instruction input for the first commodity evaluation data may be acquired in addition to the evaluation of the conventional text, picture or video. With continued reference to fig. 3, the commodity to be evaluated may be simply and quickly evaluated in a "praise" or "stepping" manner, so that not only the personal evaluation flow of the user can be simplified, but also the quality of the whole commodity can be effectively evaluated. It should be appreciated that, in this embodiment, the first evaluation instruction may be an evaluation input of "praise" or "step on" by the user.
And 206, displaying the first sub-evaluation corresponding to the first evaluation instruction at a second preset position in the evaluation area of the object to be evaluated.
FIG. 5 is a schematic diagram of one possible commodity assessment interface in the embodiment of FIG. 4. As shown in fig. 5, when evaluating the commodity to be evaluated, the data generating method provided by the embodiment can perform simple operations of "praise" or "step on" on the core evaluation, and can also input specific evaluation, so that the object to be evaluated can be evaluated in more dimensions and in detail, and the reliability and value of the commodity evaluation can be established.
And, a first sub-evaluation corresponding to the first evaluation instruction may be displayed at a second preset position in the evaluation area of the commodity to be evaluated, where the first sub-evaluation is used for evaluating the first commodity evaluation data, and the second preset position and the first preset position are adjacently disposed in the evaluation area.
Fig. 6 is a schematic diagram of a training process of a preset neural network model according to an exemplary embodiment of the present invention. As shown in fig. 6, in any of the above embodiments, the training process for the preset neural network model includes:
step 301, acquiring historical evaluation data of an object to be evaluated.
The historical evaluation data of the commodity to be evaluated can be acquired through the information acquisition interface, wherein the historical evaluation data can be summarized into a plurality of evaluation data sets.
Step 302, preprocessing the historical evaluation data.
After the historical evaluation data of the commodity to be evaluated is obtained, the historical evaluation data can be preprocessed, for example, stop words and punctuation marks in the historical evaluation data are removed.
Step 303, word segmentation processing is carried out on the historical evaluation data according to a preset word stock.
Then, word segmentation processing can be performed on the historical evaluation data according to a preset word stock, wherein the preset word stock can be a bargain word stock or other word stocks, and an expansion word stock of a custom dictionary can be added on the basis of the existing word stock.
And 304, extracting evaluation keywords of the historical evaluation data.
And 305, vectorizing the evaluation keywords.
The weight value of each word in the word segmentation set can be calculated according to a preset keyword weight algorithm, such as a TF-IDF algorithm, and the word with the weight value larger than the preset weight value is selected as an evaluation keyword of the historical evaluation data. And after the evaluation keywords are determined, vectorizing the evaluation keywords. The vectorization may be performed by Word2vec, specifically, word2vec is a group of correlation models used to generate Word vectors. These models are shallow, bi-layer neural networks that are used to train to reconstruct linguistic word text. The network is represented by words and guesses the input words in adjacent positions, and the order of the words is unimportant under the word bag model assumption in word2 vec. After training is completed, word2vec models can be used to map each word to a vector that can be used to represent word-to-word relationships, which is the hidden layer of the neural network.
Step 306, vectorizing each segmented word in the segmented word set.
In addition, vectorization can be performed on each Word in the Word segmentation set, and the specific vectorization mode can be by means of Word2vec.
Step 307, obtaining the vectorized data matrix.
And after vectorizing the evaluation keywords and vectorizing each word in the word segmentation set, constructing vectorized data into a data matrix.
Step 308, inputting the training data to the long-term and short-term memory network model for training.
And inputting the constructed data matrix into a long-term and short-term memory network model for training.
Fig. 7 is a schematic diagram of a data generating apparatus according to an exemplary embodiment of the present invention. As shown in fig. 7, the data generating apparatus provided in this embodiment includes:
an acquisition module 401, configured to acquire historical evaluation data of an object to be evaluated;
an extraction module 402, configured to extract an evaluation keyword of the historical evaluation data, where the evaluation keyword is used to characterize a physical characteristic of the object to be evaluated;
a processing module 403, configured to generate evaluation data corresponding to the evaluation keyword according to the evaluation keyword and a preset neural network model;
And the display module 404 is configured to display the evaluation data at a first preset position in the evaluation area of the object to be evaluated, so as to evaluate the object to be evaluated.
In one possible design, the obtaining module 401 is further configured to obtain a first evaluation instruction input for first evaluation data, where the evaluation data includes the first evaluation data;
The display module 404 is further configured to display a first sub-evaluation corresponding to the first evaluation instruction at a second preset position in the evaluation area of the object to be evaluated, where the first sub-evaluation is used to evaluate the first evaluation data, and the second preset position and the first preset position are adjacently disposed in the evaluation area.
In one possible design, the extraction module 402 is specifically configured to:
Performing word segmentation processing on the historical evaluation data according to a preset word stock to obtain a word segmentation set;
Calculating a weight value of each word in the word segmentation set according to a preset keyword weight algorithm;
selecting the segmentation word with the weight value larger than a preset weight value as a first evaluation keyword of the historical evaluation data, wherein the evaluation keyword comprises the first evaluation keyword.
In one possible design, the obtaining module 401 is specifically configured to:
and when the number of the items of the evaluation corresponding to the object to be evaluated is larger than a preset threshold value of the number of the items, acquiring the current commodity evaluation as the historical evaluation data corresponding to the object to be evaluated.
In one possible design, the processing module 403 is further configured to pre-process the historical evaluation data, where the pre-process includes removing stop words and punctuation marks from the historical evaluation data.
In one possible design, the extraction module 402 is specifically configured to:
and when the evaluation item corresponding to the object to be evaluated is smaller than or equal to the preset item threshold value, acquiring an evaluation keyword corresponding to the three-level category corresponding to the object to be evaluated as a second evaluation keyword, wherein the evaluation keyword comprises the second evaluation keyword.
In one possible design, the processing module 403 is specifically configured to:
generating first evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
generating second evaluation data corresponding to the object to be evaluated according to the first evaluation keywords and a preset neural network model;
Wherein the evaluation data includes the first evaluation data and the second evaluation data.
Fig. 8 is a schematic diagram showing the structure of a data generating apparatus according to another exemplary embodiment of the present invention on the basis of the embodiment shown in fig. 7. As shown in fig. 8, the data generating apparatus provided in this embodiment further includes:
the training module 405 is configured to train the preset neural network model according to the evaluation keyword and the historical evaluation data, where the preset neural network model is a long-term and short-term memory network model.
In one possible design, the training module 405 is specifically configured to:
Vectorizing the evaluation keywords and each word in the word segmentation set, and inputting the vectorized data matrix as training data to train the long-term and short-term memory network model.
The above processing module 403 may be configured as one or more integrated circuits implementing the above methods, for example, one or more Application SPECIFIC INTEGRATED Circuits (ASIC), or one or more microprocessors (DIGITAL SINGNAL processor, DSP), or one or more field programmable gate arrays (Field Programmable GATE ARRAY, FPGA), or the like. For another example, when a module above is implemented in the form of a processing element scheduler code, the processing element may be a general-purpose processor, such as a central processing unit (Central Processing Unit, CPU) or other processor that may invoke the program code. For another example, the modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in hardware plus software functional units. The data generating device provided in the embodiments shown in fig. 7 to fig. 8 may be used to execute the data generating method provided in any of the above method embodiments, and the specific implementation manner and technical effect are similar, and are not repeated here.
Fig. 9 is a schematic structural view of an electronic device according to an exemplary embodiment of the present invention. As shown in fig. 9, an electronic device 500 provided in this embodiment includes:
processor 501, and
A memory 502 for storing executable instructions of the processor, which may also be a flash memory;
Wherein the processor 501 is configured to perform the steps of the above-described method via execution of the executable instructions. Reference may be made in particular to the description of the embodiments of the method described above.
Alternatively, the memory 502 may be separate or integrated with the processor 501.
When the memory 502 is a device separate from the processor 501, the electronic device may further include:
A bus 503 for connecting the processor 501 and the memory 502.
The present embodiment also provides a readable storage medium having a computer program stored therein, which when executed by at least one processor of an electronic device, performs the methods provided by the various embodiments described above.
The present embodiment also provides a program product comprising a computer program stored in a readable storage medium. The computer program may be read from a readable storage medium by at least one processor of an electronic device, and executed by the at least one processor, causes the electronic device to implement the methods provided by the various embodiments described above.
Those of ordinary skill in the art will appreciate that all or a portion of the steps of implementing the various method embodiments described above may be implemented by hardware associated with program instructions. The foregoing program may be stored in a computer readable storage medium. The program, when executed, performs the steps comprising the method embodiments described above, and the storage medium described above includes various media capable of storing program code, such as ROM, RAM, magnetic or optical disk.
It should be noted that the above embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the technical solution described in the above embodiments may be modified or some or all of the technical features may be equivalently replaced, and these modifications or substitutions do not make the essence of the corresponding technical solution deviate from the scope of the technical solution of the embodiments of the present invention.