As part of the national innovation system, science-tech journals are currently gaining importance. The development and progress of subjects are directly related to the quality of journals. A scientific and systematic evaluation of journals can help researchers understand the current research priorities and hot spots, as well as improve their quality and influence. Sci-tech journals may benefit greatly from the creation of a dynamic evaluation model for journals based on journal metrics and time series data, as well as its rationality and usability.diaries and develop another time series dataset of 18 diary assessment measurements. In light of this, we developed nine models in the field of time series analysis using standard machine learning and deep learning techniques. We then trained and experimented with the dataset to produce extensive and varied evaluation metric results, compared and analyzed the results to confirm the generalizability of these techniques for the comprehensive dynamic evaluation of journals, and discovered that the LSTM model we developed achieved high prediction accuracy in the task proposed in this paper, laying the groundwork for subsequent research on this issue.

Limitations. We gathered data on journal evaluation metrics from the Wanfang and ZhiWang platforms over the past five years to create the dataset. However, there are still numerous journal platforms and search websites that can provide additional data or proofread existing data. We

assembled models utilizing nine standard time series anticipating techniques to powerfullyassess sci-tech diaries, and albeit these models are general for the assignment of dynamicassessment of sci-tech diaries, they have not been explicitly streamlined for ourmultivariate brief time frame multi-series dataset. Additionally, these limitations provide directions for future research.

Future research More journal metric time series data from various data sources could be added to the models built in this study for dynamic journal evaluation to boost their accuracy and credibility. We might think about selecting some covariates for model training based on how important the journal metrics are from the perspective of the feature value. In terms of the algorithm, we could think about adding a transformer or a Seq2Seq structure, among other things, to the network structure and parameter settings of the model built using LSTM and GRU methods in this study.

Implications in Practice This study's model can be used as a guide for resource allocation and subsequent journal development planning, and it can be extended with a variety of application strategies for future journal evaluation and analysis tasks. The model developed in this study, for instance, can be used to outline the possible future development status of journals when attempting to make decisions regarding the allocation of journal resources. This can help guide the reasonable allocation of journal resources to some extent. For example, when there are certain arranging assumptions for the improvement of diaries in a given year, they can be passed into the models as contribution to acquire dynamic assessment results, and unique dynamic assessment results can be acquired through the nonstop change of the arranging, which can, thusly, help with changing diaries' turn of events.Implications From the Theory The extensive results of this study's experiments can serve as methodological references for other studies of multivariate short-time multi-series time series analysis, and the dataset proposed in this paper can also support similar endeavors.

Regards

David William

Manager Editor

Advance in Applied  Science