A method and system for predicting hot journal topics

Abstract

The application provides a hot journal topic prediction method and system, which comprises the following steps: sampling and crawling literatures in a period, calculating the heat of each topic word in each period, and predicting the heat of the topic word in a future period based on a time series model; obtaining a topic word weight vector of a paper of a journal and calculating the influence value of the paper in a period; establishing the connection between the predicted heat of the topic word in a period and the influence value of the paper in a period by using a prediction model, predicting the influence of different topics in a future period through the heat of the topic word, and recommending a journal topic based on the predicted heat of the topic word in a future period and the prediction model.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of computer application, and particularly relates to a hot journal topic prediction method and system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] Optimizing journal influence indicators is an important measure to improve journal publishing quality. Journal influence indicators usually include the number of submissions, the number of paper readings, the number of citations, the number of downloads, etc. Papers of different topics have different contributions to the influence of journals. If the topic of a paper is a current research hotspot, the paper will have a high influence in the future period after being published, i.e. high reading volume, high download volume, high citation volume, etc. For example, when the excellent machine learning model deep learning technology was just launched, the related theoretical and application papers were read by more scholars.

[0004] From the above phenomenon, it can be seen that publishing papers on topics related to research hotspots can effectively improve the influence of journals. Journal editors can conduct a special issue, a special issue, or a supplement to solicit papers on specific topics. However, it is a very difficult and time-consuming task to select the current hot topic that meets the scope of journal publishing from numerous research topics.

[0005] In the rapid development of society, a large amount of information and data are generated. In recent years, with the improvement of computer hardware performance and the continuous optimization of data analysis algorithms, people's ability to record and mine data is becoming stronger and stronger. It is possible to find and use truly valuable information from large-scale complex data. Big data has attracted attention and attention in various fields. Using big data to assist journal topic selection is inevitable with the development of technology and an effective tool for journal operators.

[0006] At present, the determination of the scope of journal publishing topics is mainly based on subjective judgment, and there is a problem of insufficient prediction accuracy. In addition, there are existing technologies for predicting hot topics in related disciplines and predicting hot topics in technical fields. However, the above technologies are mainly for predicting hot topics in the technical field, and the hot topics of journals have the characteristics of suddenness and real-time, while the hot topics in the technical field have the characteristics of path dependence and demand orientation. The characteristics of journal topic hotspots and technical field hotspots are obviously different, and the above hot topic prediction technologies in the technical field cannot be applied to hot journal topic prediction. SUMMARY

[0007] In order to overcome the shortcomings of the above prior art, the present application provides a hot journal topic prediction method, which predicts hot journal topics in the future period based on currently available information.

[0008] To achieve the above object, one or more embodiments of the present application provide the following technical solutions:

[0009] In a first aspect, a hotspot journal theme prediction method is disclosed, comprising:

[0010] Sampling and crawling documents of a period, calculating the heat of each topic word in each period, and predicting the heat of topic words in a future period based on a time series model;

[0011] Obtaining the topic word weight vector of a journal paper and calculating its influence value in a period;

[0012] Using a prediction model to establish the connection between the predicted heat of topic words in a period and the influence value of papers in a period, and predicting the influence of different topics in a future period through the heat of topic words;

[0013] Based on the predicted heat of topic words in a future period and the prediction model, recommending journal themes.

[0014] As a further technical solution, when sampling and crawling documents of a period, the crawler technology is used to sample and crawl documents of a period in the literature database, and the crawled articles are texted to obtain a set of science and technology information articles collected in a period , wherein represents the cycle number.

[0015] As a further technical solution, the heat of each topic word in each period is calculated, and for period, the calculation steps are:

[0016] Obtain the topic word weight vector of a document in the set of all documents of period;

[0017] Record the download volume and citation volume of the document in period;

[0018] Obtain the search volume of each topic word in the search engine through the crawler technology;

[0019] Based on the topic word weight vector, the download volume and the citation volume, and the search volume, calculate the heat of the topic word component in the word bank in period.

[0020] As a further technical solution, when predicting the heat of topic words in a future period, it comprises:

[0021] Integrate the heat of the topic word component in the word bank in each period in period to form a time series of the heat of each topic word;

[0022] predicting the popularity of the subject word in the future period of time through the time series prediction model for each subject word.

[0023] As a further technical solution, the subject word weight vector of a paper of a journal is obtained, specifically:

[0024] obtaining the text weight vector and the keyword weight vector of a paper of a journal;

[0025] calculating the subject word weight vector of the set of cited documents of a paper of a journal;

[0026] calculating the subject word weight vector of a paper of a journal.

[0027] As a further technical solution, the subject word weight vector of the set of cited documents of a paper of a journal is calculated, specifically:

[0028] calculating the text weight vector of each cited document;

[0029] each cited document is multiplied by the number of citations to obtain the arithmetic mean of each text vector;

[0030] forming the text weight vector of the cited document.

[0031] As a further technical solution, in the prediction model, each subject word corresponds to a sub-network, all sub-networks are called subject word sub-networks, and the network is not connected to other sub-networks.

[0032] The input of the first node of the sub-network is the popularity value of the subject word, and the second node is the weight vector of the subject word in the paper;

[0033] When the subject weight vector is greater than or equal to the set threshold value, the corresponding sub-network is connected to the main network;

[0034] When the subject weight vector is less than the set threshold value, the sub-network is deleted from the main network.

[0035] In a second aspect, a hot topic journal subject prediction system is disclosed, comprising:

[0036] The hot topic analysis module is configured to: sample and crawl documents in a period of time, calculate the popularity of each subject word in each period of time, and predict the popularity of the subject word in the future period of time based on a time series model;

[0037] The paper subject analysis module is configured to: obtain the subject word weight vector of a paper of a journal and calculate the influence value thereof in a period of time;

[0038] The prediction model training module is configured to establish the relationship between the hotness of the subject word in the predicted period of time and the influence value of the paper in the period of time by using the prediction model, and predict the influence of different subjects in the future period of time through the hotness of the subject word.

[0039] The journal topic recommendation module is configured to recommend the journal topic based on the predicted hotness of the subject word in the future period of time and the prediction model.

[0040] The above one or more technical solutions have the following beneficial effects:

[0041] The present application researches the hot spot analysis, the paper theme analysis, the prediction model training and the like, realizes the topic selection of the special journal and the special topic, and makes the selected theme more in line with the current hot spot.

[0042] The technical scheme of the present application can also predict the size of the journal influence of the submitted paper, and facilitate the journal editor to select the priority accepted manuscript.

[0043] The advantages of the additional aspects of the present application will be partially given in the following description, partially become obvious from the following description, or be known by the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0044] The drawings accompanying the specification of the present application form a part thereof, serve to provide further understanding of the present application, and together with the description of the exemplary embodiments of the present application and the explanation thereof serve to explain the present application, and do not constitute improper limitations on the present application.

[0045] Figure 1 The module structure diagram of the topic planning auxiliary decision-making system of the embodiment of the present application;

[0046] Figure 2 The structure schematic diagram of the prediction model of the embodiment of the present application;

[0047] Figure 3 The process schematic diagram of the prediction model of the embodiment of the present application. DETAILED DESCRIPTION

[0048] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0049] It should be noted that the terms used herein are only for the purpose of describing the specific embodiments, and are not intended to limit the exemplary embodiments according to the present application.

[0050] In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0051] Embodiment one

[0052] The embodiment discloses a hot journal topic prediction method, comprising:

[0053] Step one: sample and crawl the literature of a period, calculate the heat of each topic word of each period, and predict the heat of the topic word in the future period based on a time series model;

[0054] Step two: obtain the topic word weight vector of the papers of a journal and calculate its influence value in a period;

[0055] Step three: use the prediction model to establish the relationship between the heat of the predicted topic word in a period and the influence value of the papers in a period, and predict the influence of different topics in the future period through the heat of the topic word;

[0056] Step four: recommend the journal topic based on the predicted heat of the topic word in the future period and the prediction model.

[0057] In the above step one, specifically:

[0058] 1-1) Use crawler technology to sample and crawl the literature of a period in the literature database. Based on OCR technology, the crawled articles are texted, and the set of science and technology information articles grabbed in a period is denoted as , wherein represents the cycle number.

[0059] 1-2) Calculate the heat of each topic word of each period. For cycle, the topic word is calculated based on the following method:

[0060] A) Let be the set of all literature of cycle, and use to identify a piece of literature. Based on the TF-IDF algorithm, the topic weight vector of is obtained, denoted as , is the set of words in the science and technology vocabulary, is the corresponding topic word component in the vocabulary , and the obtained weight TF-IDF value is denoted as , wherein is the literature identifier, is the word identifier.

[0061] B) For , record its download amount and citation amount in cycle.

[0062] C) Obtain each topic word by crawler technology In search engine search volume, set .

[0063] D) Set is In Period heat. The calculation method is as follows:

[0064]

[0065] Among them are The average value. This method considers the influence of download volume, citation volume and search volume on heat.

[0066] 1-3) Predict the heat of topic words in the future period. Integrate each period , and form the time series of the heat of each topic word. Set as Heat time series. Set as the current period, for each , the heat of can be predicted by the time series prediction model.

[0067] The specific method is as follows: first, take number As training data; second, select advanced time series models such as LSTM, ARIMA, etc.; Finally, predict the heat of each topic word in .

[0068] The above step two is specifically:

[0069] The main tasks of the paper theme analysis are: one is to analyze the weight of the paper theme word; Two is to analyze the contribution of the paper to the influence of the journal in a period of time.

[0070] For a journal paper, the theme word weight can be analyzed as follows:

[0071] 2-1) Set is a paper of the journal. Based on TF-IDF algorithm, the text weight vector of is obtained, denoted as . Based on TF-IDF algorithm, the keyword weight vector of is obtained, denoted as .

[0072] 2-2) Set is the citation literature set of . Calculate The keyword weight vector is calculated as follows: First, the word weight vector of each cited document is calculated based on TF-IDF; second, each word vector is set as... , It is obtained by multiplying the vector weights of each cited reference by the number of citations and then taking the arithmetic mean; finally, the word weight vector of the cited references is formed, denoted as... .

[0073] 2-3) Papers The keyword weight vector is calculated as follows: , .in , , These are the coefficients of the three variables.

[0074] For a paper published in a journal, its contribution to the journal is quantified using an impact value. The impact value is calculated over a period of time as follows:

[0075] set up Influence value ,but ,in , , They are The number of reads, downloads, and citations within a certain period of time. , , The coefficients for the three variables are specified separately. The coefficients can be set according to the variable's contribution to the influence.

[0076] Step three above specifically includes:

[0077] Having obtained data on research hotspots and paper impact, a predictive model can be used to establish a relationship between the two, that is, to predict the impact of different topics over a future period using research hotspot data. The steps are as follows:

[0078] 3-1) Organize training data. A paper published when the topic is a hot research area will have a higher impact. This involves compiling a period of training data. and the keyword weights of each paper. As input values, the impact value of each paper will be used in the next few cycles. As output value.

[0079] That is:

[0080] .

[0081] in, for Influence value For a collection of papers, as the article, as the paper 's subject weight vector, as the set of words in the science dictionary, is the keyword in periodic heat

[0082] 3-2) Prediction model structure design. Based on the improved deep neural network to predict the influence of papers, the model structure is as follows Figure 2 . The improved deep neural network model is as follows: each subject word corresponds to a subnetwork, which is called a subject word subnetwork. The subnetwork is not connected to other subnetworks. The input of the first node of the subnetwork is the heat value of the subject word , and the second node is the weight vector of the subject word in the paper . All subnetworks are connected to the main network. Excessive network connection will cause difficulty in training. To simplify the model structure, selective connection can be set. When the subject weight vector is greater than or equal to the set threshold, the connection is maintained; when the subject weight vector is less than the set threshold, the connection is deleted. The above improved deep neural network can effectively control the network size and improve the training speed. This network can dynamically adapt to different hot words, improve the accuracy of training and prediction.

[0083] 3-3) Prediction model training process. There are many kinds of papers, such as review, brief report, basic research, applied research, etc. The influence of each kind of paper is different from the correlation degree of research hotspots, so separate models are used for independent training of different types of papers.

[0084] As Figure 3 , the sample library is divided into different models according to the type of paper. Set as a trained model, as a model set. Set as the prediction output value.

[0085] The above step four is specifically:

[0086] 4-1) Predict hot topics. Journal special topics publish group information to the paper published, which needs to go through a long time. Therefore, the special topic selection should have a certain predictability. Set the journal publication date as , predict the first hot topic words, which form the hot topic word set .

[0087] 4-2) Define the subject correlation coefficient , represents two subject words , The possibility of co-occurrence and correlation in all papers of a journal. The following formula is used to calculate:

[0088] (1)

[0089] where, represents a paper, represents the set of all papers of the journal; represents the weight of a subject word in , represents the weight of a subject word in ; represents the average weight of a subject word in , represents the average weight of a subject word in .

[0090] 4-3) For each , calculate its correlation coefficient with other subject words based on formula (1).

[0091] 4-4) For each , generate a vector , where .

[0092] 4-5) For each , input into each model of , and calculate . . Where, is the heat value of the word predicted by module 1 for the period.

[0093] 4-6) The maximum in step 4-5) is the recommended topic word, where the corresponding paper type represented by is the recommended type of paper for publication.

[0094] Through the above steps, the hottest research topic at the time of publication can be recommended. The type of article that can be recommended for publication can be recommended. Based on module 1, there is a certain prediction accuracy.

[0095] More specifically, taking a certain power industry journal as an example, the journal plans to launch a special issue related to power in six months. The process of assisting in its topic selection is as follows:

[0096] 1) Analyze current research hotspots based on the research hotspot analysis module. Predict the top 100 research hotspot keywords six months from now, such as "deep learning" and "vaccine".

[0097] 2) Analyze the relevance of these terms to historically published papers in power industry journals within the scientific and technological terminology database, i.e., the probability of them appearing together in papers. The analysis shows that research topics in unrelated fields, such as "vaccines," have low relevance; while research topics in related fields, such as "flexible power transmission," show high relevance. Furthermore, calculations are performed based on these relevances. .

[0098] 3) Heat value and Input into the prediction model and select the largest value. Therefore, the keyword "flexible power transmission" has the greatest [potential]. Therefore, we recommend "flexible power transmission" as a topic for the power industry special issue six months from now.

[0099] 4) It was also found that applied research papers and research review papers have a higher impact, so it is recommended that journal editors give priority to accepting these two types of papers.

[0100] Example 2

[0101] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0102] Example 3

[0103] The purpose of this embodiment is to provide a computer-readable storage medium.

[0104] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above method.

[0105] Example 4

[0106] See appendix Figure 1 As shown, the purpose of this embodiment is to provide a hot topic prediction system for journals, including:

[0107] The research hotspot analysis module is configured to: sample and crawl literature for a period of time, calculate the popularity of each topic word for each period of time, and predict the popularity of topic words for a future period of time based on a time series model;

[0108] The main function of the module is to record the heat of the research topic in the past period of time and the current research topic, and predict the heat of the research topic in the future period of time. The module provides training data for the prediction model training module and provides prediction data for the journal topic recommendation module.

[0109] The paper topic analysis module is configured to obtain a topic word weight vector of a paper of a journal and calculate an influence value of the paper in a period of time.

[0110] The main function of the above module is to analyze the topic involved in the paper and collect the influence index of the paper. The module provides training data for the prediction model training module.

[0111] The prediction model training module is configured to use a prediction model to establish the relationship between the heat of the topic word predicted in a period of time and the influence value of the paper in a period of time, and predict the influence of different topics in the future period of time through the heat of the topic word.

[0112] The main function of the above module is to train the topic prediction model. After the model training is completed, the model is sent to the journal topic recommendation module.

[0113] The journal topic recommendation module is configured to recommend the journal topic based on the predicted heat of the topic word in the future period of time and the prediction model.

[0114] The main function of the above module is to recommend the topic of the journal.

[0115] The technical scheme of the present application is based on the currently available information such as literature data, social news, etc., and designs a feasible and effective prediction model to provide suggestions for the journal topic in the future period of time.

[0116] The steps and methods involved in the devices of the above embodiments two, three and four correspond to the method embodiment one, and the specific implementation can be referred to the related description part of the embodiment one. The term "computer readable storage medium" should be understood as including a single medium or multiple media of one or more instruction sets; it should also be understood as including any medium capable of storing, encoding or carrying instruction sets for execution by a processor and causing the processor to perform any method in the present application.

[0117] Those skilled in the art should understand that each module or each step of the present application described above can be realized by a general computer device, alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be respectively manufactured into each integrated circuit module, or a plurality of modules or steps among them can be manufactured into a single integrated circuit module to realize. The present application is not limited to any specific combination of hardware and software.

[0118] The above describes the specific embodiments of the present application in combination with the drawings, but is not a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. A method for predicting a topic of a hot journal, characterized by, The method comprises the following steps: sampling and crawling literatures in a period of time, calculating the heat of each topic word in each period, and predicting the heat of the topic word in a future period of time based on a time series model; The calculation of the heat of each topic word in each period is performed for The calculation step is: obtaining a subject heading weight vector for a document in the collection of all documents of the cycle; records its in periodic download and citation volume; obtaining the search volume of each topic word in a search engine through a crawler technology; Based on the theme word weight vector, download and reference amount and search amount, the theme word contribution in the word library is calculated Periodicity obtaining the topic word weight vector of a paper of a journal and calculating the influence value of the paper in a period of time; establishing the correlation between the predicted heat of the topic word in a future period of time and the influence value of the paper in a period of time by using a prediction model, and predicting the influence of different topics in a future period of time through the heat of the topic word; in the prediction model, each topic word corresponds to a sub-network, all sub-networks are called topic word sub-networks, and the network is not connected with other sub-networks; the input of the first node of the sub-network is the heat value of the topic word, and the second node is the weight vector of the topic word in the paper; when the topic weight vector is greater than or equal to a set threshold, the corresponding sub-network is connected to the main network; when the topic weight vector is less than the set threshold, the sub-network is deleted from the main network; based on the predicted heat of the topic word in a future period of time and the prediction model, the theme of the journal is recommended.

2. The method of claim 1, wherein the step of predicting the topic of the hotspot periodical is characterized by, The sampling crawling one period of literature, using crawler technology in literature database sampling crawling one period of literature, the crawled article text, get a set of science and technology information articles in the collection of articles in a period of time Wherein Indicates the cycle number.

3. The method of claim 1, wherein the step of predicting the topic of the hotspot periodical is characterized by, The method for predicting the heat of the topic word in a future period of time comprises the following steps: The subject word vocabulary components in the vocabulary of each cycle are integrated to form a time sequence of the subject word heat of each subject word. The heat of each cycle is integrated to form a time sequence of the heat of each subject word. for each topic word, the heat of the topic word in a future period of time is predicted through a time series prediction model.

4. The method of claim 1, wherein the step of predicting the topic of the hotspot periodical is characterized by, The method for obtaining the topic word weight vector of a paper of a journal comprises the following steps: obtaining the text weight vector and the keyword weight vector of a paper of a journal; calculating the topic word weight vector of the citation literature set of the paper of the journal; calculating the topic word weight vector of the paper of the journal.

5. The method of claim 4, wherein the step of predicting the topic of the hotspot periodical is characterized by, The method for calculating the topic word weight vector of the citation literature set of the paper of the journal comprises the following steps: calculating the weight vector of each citation literature; each citation literature weight vector is multiplied by the number of citations and then arithmetically averaged to obtain each weight vector; forming the weight vector of the citation literature.

6. A hot journal topic prediction system employing the method of claim 1, wherein, The method comprises the following steps: a research hotspot analysis module is configured to sample and crawl literatures in a period of time, calculate the heat of each topic word in each period, and predict the heat of the topic word in a future period of time based on a time series model; a paper topic analysis module is configured to obtain the topic word weight vector of a paper of a journal and calculate the influence value of the paper in a period of time; a prediction model training module is configured to establish the correlation between the predicted heat of the topic word in a future period of time and the influence value of the paper in a period of time by using a prediction model, and predict the influence of different topics in a future period of time through the heat of the topic word; a journal topic recommendation module is configured to recommend the theme of the journal based on the predicted heat of the topic word in a future period of time and the prediction model.

7. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method of any one of claims 1-5.

8. A computer readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to perform the steps of the method of any one of claims 1-5.

Images