Research on Resource Sharing Methods of English Translation Corpus in Colleges and Universities under the Background of Informatization

One of the characteristics of the translation corpus is that the corpus material is very rich, which not only has a large number of theoretical and professional materials, but also has a large number of humanized and targeted training materials that are suitable for the students’ learning, of course, standard materials and rules and regulations are also a part of the material [1-3]. The material of translation corpus is based on the actual learning of college students in daily English professional translation courses, and its corpus material system can meet the students’ needs for the richness and humanized application of corpus resources [4-6].

From an objective point of view, the corpus material can reflect that the vocabulary ecological context it provides to the students is real and effective, so English teachers can utilize the corresponding contextual vocabulary and carry out vocabulary exploration from two different perspectives, both vertically and horizontally, when letting students study English translation based on the translation corpus [7-9]. Vocabulary is a very important element in English translation, and it is also an important element for students to learn English professional translation courses, when teachers teach students the corresponding vocabulary search and application methods based on the translation corpus materials, students can independently utilize the translation corpus to continuously exercise their own translation ability [10-13].

The vivid contexts in the translation corpus can also reflect real translation application scenes, and different scenes have different contextual themes, so teachers can guide students to explore their own learning weaknesses and practical problems, and use the translation corpus to make targeted training and breakthroughs, so as to have a deeper understanding of the relevant translation methods and techniques [14-16]. Therefore, it is necessary to build and improve the quality of corpus construction, and at the same time realize the dynamic and open expansion of the corpus in order to promote the updating, capacity and representativeness of the shared corpus and other advantages can be improved [17-18].

Scholars’ studies on the establishment and optimization of English language corpora are as follows, literature [19] describes the Multilingual Student Translation (MUST) corpus, which was built by an international team of language translators and language learners, and is characterized by the fact that each text consists of a rich set of standardized metadata related to the corresponding source text, the translation task, and the student. Literature [20] dialectically views the research related to corpus translation and argues that these limitations can be overcome by adopting a revised research agenda, employing a richer research methodology and more theoretical awareness to develop empirical research on corpus translation. Literature [21] builds a multilingual speech translation corpus that can effectively assist end-to-end language translation teaching and training. Literature [22] designed a PSO-BP model based on bp neural network and particle swarm optimization algorithm, which significantly improves the prediction accuracy of BPNN, and finally looked forward to the integration of blockchain technology into the English translation corpus application path. Literature [23] analyzes the connotation and technical logic of corpus in detail and focuses on the alignment technology of corpus, and based on this, it thinks about the hybrid English translation teaching mode based on corpus English translation. Literature [24] envisioned a framework with corpus management subsystem and query statistics subsystem as the infrastructure, which was introduced into English translation teaching in colleges and universities, enhancing the openness of English translation teaching and promoting the development of students’ cross-cultural awareness and thinking. The above research mainly focuses on the path research of corpus in English translation teaching, and involves the introduction of corpus-based resource sharing for English translation teaching.

With the goal of realizing the sharing of English translation corpus resources in colleges and universities, this paper proposes a cloud platform for English translation corpus resources in combination with cloud computing technology, which promotes the sharing of English translation corpus resources to become possible. The overall architecture of the cloud platform is designed at four levels, namely, infrastructure layer, application interface layer, educational application layer and user interface layer, and corresponding algorithms are proposed for recommending and sharing resources of English translation corpus to realize the resource recommendation and sharing functions of the cloud platform. In terms of the resource recommendation function, the online resource recommendation model is constructed based on the immuno-evolutionary algorithm, the recommendation weights are updated according to the immuno-evolutionary adaptation relationship, the search space is generated and the immuno-evolutionary resource recommendation formula is calculated. Based on the user feedback data, the search space is generated, the local optimal solution is obtained through immuno-evolutionary computation, and the global optimization search is carried out quickly to design the online teaching resources recommendation engine. As for resource sharing, the features of teaching resources are extracted and a table of teaching resource word adjacency matrix is established. Based on the extracted features of candidate feature words in teaching resources to construct classification criteria, complete the classification of resources, and then simplify the degree of similarity and difference of resources, improve the circulation between different resources, and realize the sharing of teaching resources to the maximum extent. Finally, the virtualization technology means based on KVM is used to install QEMU-KVM virtual machine, virtual management suite and other devices to finally realize the resource sharing ability of the cloud platform. For the resource recommendation and sharing core functions of the English translation corpus resource cloud platform in this paper, recommendation performance experiments and sharing performance experiments are carried out respectively to check the effectiveness of the functions.

2

English Translation Corpus Resource Cloud Platform

With the development of the pace of educational informatization, colleges and universities have invested a lot of human and material resources in the construction, sharing and management of software and hardware of English teaching resources, among which the teaching resources of English translation corpus have also shown a geometric growth. However, the traditional information technology to realize the English translation corpus resources resource sharing to a certain extent is blocked and difficult to develop. This paper aims to realize the sharing of English translation corpus resources in colleges and universities, combines the advantages and characteristics of cloud computing, and builds a cloud platform for English translation corpus resources.

2.1

Cloud Computing Architecture

Cloud computing is a computing model that provides dynamically scalable virtualized resources as a service over the Internet [25]. Cloud architecture can be understood as the integration and delivery of hardware and software resources, the integration of hardware and software resources in the cloud through the technology of the architecture, and finally provided to the user through the cloud service mode. The common understanding of the IT industry for the cloud architecture system is to divide it into three layers, which are the Infrastructure Layer (IaaS), the Platform Layer (PaaS), and the Application Layer (SaaS). 1)

Infrastructure Layer

The main realization is the service on the infrastructure level, using virtualization technology to structure the infrastructure layer and then dynamically deliver the abstracted underlying resources to the demanders outside the cloud system, so as to flexibly and conveniently realize on-demand service provision.

2)

Platform Layer

It is actually a scalable, secure and highly available cloud service environment that can be delivered for development, in which various conditions required for development are gathered, and the central function is to manage software resources.

3)

Application Layer

Various application software corresponding to business requirements are bundled and run on the platform cloud according to standardized business logic, and users interact with the services provided in the cloud through the application layer.

2.2

Overall Architecture Design

The English translation corpus resource cloud platform built in this paper is specified as follows. 1)

Infrastructure Layer

The infrastructure layer is the lowest layer of the cloud architecture, and through virtualization technology, it can realize the upward layer to give resources such as storage, network and computing power, etc. The infrastructure layer consists of two parts, namely, the physical hardware sub-layer and the virtualization sub-layer. Physical hardware sub-layer is the actual existence of a variety of physical hardware (network equipment, memory, CPU, storage devices, etc.). The virtualization sub-layer, which is built on top of it using virtualization technology, is used to manage various physical hardware devices. Therefore, the virtualization sub-layer is actually composed of various types of virtual machines, which provide virtual computing resources, storage space and network communication capabilities and other resources for the upper layer.

2)

Application Interface Layer

Provides PaaS, the upper layer built on the basis of infrastructure layer (physical hardware sub-layer + virtualization sub-layer), which is used to provide programmers with common APIs and development tools. Programmers can obtain interfaces in the form of Web services through the public API to realize the development of different kinds of cloud services, such as teaching software, etc., and they can also develop their own private Web services through the public API, which reduces a lot of trouble in the process of system management or platform construction for the majority of programmers in the process of realization.

3)

Education Application Layer

Providing SaaS, the educational application layer is characterized by the fact that the application software provided are all constructed for education and teaching, and the application layer is implemented with education and teaching as the leading direction.

4)

User Interface Layer

In order to realize that users can easily access the remote education cloud from the client side, the user interface layer provides the traditional graphical user interface (GUI) and the current mainstream browser-based Web interface in two ways. The browser-based Web interface relies on Web2.0 technology, which is the mainstream of the current user interface layer.

2.3

Platform Functional Algorithms

2.3.1

English Translation Corpus Resource Recommendation Algorithm

Immuno-evolutionary algorithm is an intelligent optimization algorithm that simulates the evolutionary process of biological immune system, which can simulate the process of antibody mutation and recombination of biological immune system, generate new solution vectors, and obtain the optimal solution [26]. At the same time, the immune evolution algorithm has a memory function, which can quickly retrieve and match the already recommended resources and accelerate the search process. Therefore, the online teaching resources recommendation model is constructed based on the immuno-evolutionary algorithm. Firstly, the global assumption is made, at this time the recommendation vector X_i can be expressed as: (1) $X_{i} = (X_{i}^{1}, X_{i}^{2}, \dots, X_{i}^{D}), i = 1, 2, 3, \dots, N$

Where $X_{i}^{1}, X_{i}^{2}, \dots X_{i}^{D}$ denotes the 1 ~ D-item resource recommendation node.

N denotes the particle dimension. The recommendation weights $V_{I}^{D}$ are updated according to the immuno-evolutionary adaptation relationship, calculated as Eq: (2) $V_{I}^{D} = ω V (k) + c (p_{B e s t} - X)$

Where ω denotes the inertia weight, V(k) denotes the objective recommendation function, c denotes the particle update speed, p_Best denotes the optimal random dimension, and X denotes the limiting speed range. When the above steps are completed, the user’s feedback data can be collected to randomly generate a set of initialized antibodies to generate the search space, at this time, the parent generation generates the immuno-evolutionary resource recommendation formula for the child generation, which is computed by the formula (3) $σ = σ_{k} + σ_{n} e^{k}$

Where σ_k denotes the number of individuals in the population, σ_n denotes the individual number of the optimal immune cell, and e^k denotes the individual standard deviation. At this point, the global search and mixing adjustment can be performed according to the normal distribution stochastic criterion, calculated as: (4) $x_{n} = σ_{k} + \frac{σ_{n} + 1}{2 (μ + k)}$

Where μ denotes the optimization search parameter and k denotes the number of iterations. Based on this, immuno-evolutionary computation can be carried out to obtain the local optimal solution, at this time the constructed teaching resource recommendation model can be expressed as: (5) $E = σ_{k} + N x_{i d}$

where N denotes the number of evolutionary generations and x_id denotes the population size. Global optimization can be performed quickly using Eqs. (1) to (5).

In the process of online English translation corpus resource recommendation, it is necessary to consider important recommendation reference indexes, respond quickly to user recommendation information, and improve the rationality of teaching resource recommendation. The article designs an online teaching resources recommendation engine based on user characteristics.

The recommendation engine can directly interact with the user and quickly receive the information lost by the user. The online teaching resources recommendation engine can also record the scoring information after the recommendation of teaching resources, and carry out operational data storage and offline computation, so as to quickly generate the online teaching resources recommendation model.

2.3.2

Algorithm for Resource Sharing in English Translation Corpus

1)

Resource Characterization Extraction

The main goal of teaching resource feature extraction is to filter and reduce the burden so that it has less impact and less information. For this reason, the article proposes a more appropriate method for feature extraction of teaching resources.

The development of growth matrix can determine the semantic relationship between resource categories and is realized through 3 stages, first, using word segmentation tool to segment the semantics of words in the text and displaying the segmented words, second, distinguishing between words in teaching resources and calculating the equivalence between words and words respectively, and third, building a table of adjacency matrices of teaching resources words based on similarity values.

W_i and W_j denote the dynamic and static names of teaching resources, respectively. On the basis of satisfying the adjacency matrix relationship of teaching resource words, the feature of the vocabulary is set to λ. When the value of $L (W_{i}, W_{j})$ is greater than or equal to the feature λ of the vocabulary, it indicates that the candidate word plays a major role in the features of the teaching resource. All W_j that satisfy the condition set $L (W_{1}, W_{2}) \geq λ$ are used as the next W_i notices. The main content and terms expressed in the teaching resources are considered as featured words F and the selection of the names of the features follows two points. First, if there is a high-level connection between the teaching resource and the discussed lexicon, it is considered a strong expression and can be selected as a feature marker. Second, in the case of a high degree of word characterization, the featured term F has a very high record of relevance to the teaching resource.

In summary, $F (W_{j})$ is the W_j feature level of the featured word F, $F (W_{j})$ as a set of composite names where the feature amount of $F (W_{j})$ is equal to the feature amount of the featured word F when the feature level of $F (W_{j})$ is all equal to the total number of conditions for $L (W_{1}, W_{2}) \geq λ$ . The value of $F (W_{i})$ is: (6) $F (W_{i}) = \sum \frac{F (W_{j})}{n j}$

Where: n is the feature quantity.

Assuming that the teaching resource has n candidate names and that a set of function vectors is represented using a n-dimensional column vector, i.e., $H_{x} = [F (W_{1}) F (W_{2}), F (W_{3}), \dots, {F (W_{n})]}^{T}$ , the function vector is obtained as: (7) $H_{x} = F (W_{i}) H_{x - 1}^{T}$

Since Eq. (7) is obtained by cyclic definition, when H_x > 1, the final result can be recalculated to obtain the final result.

When $H_{x}^{T} = [\begin{array}{l} 0 & 1 \\ 1 & 0 \end{array}]$ , H₁ = H₂, the computational degree of the candidate sequence parameters can be obtained as Eq: (8) $H_{x} = \frac{(1 - d) H}{n} + d H_{x}^{T}$

Where H is a n×n matrix and d is the value of the matrix element taking the value of [0,1], which takes the value of 0.85 in the article.

The article selects the words with higher function from H_x as candidate words to realize the feature extraction of teaching resources. 2)

Resource Classification

The features of teaching resources can be classified by constructing classification criteria based on the features of the candidate feature words in the extracted teaching resources. Teaching resources classification mainly consists of two parts.

First, the constituent features of hierarchical relationships are selected in the setup training sample W_in. The commonly used $(A_{i}, B_{j})$ represents the relationship between the upper and lower two words. The semantic tree structure is used to compute the relationship between the upper and lower features. To verify the relationship between this hierarchical concepts, the semantic distance of the first training resource is used to reflect the strength of the hierarchical relationship $(A_{i}, B_{j})$ , which is calculated by the formula: (9) $D e g r e e (A_{i}, B_{j}) = \frac{2 \partial H_{x}}{\partial + l}, 0 < \partial \leq 1$

where ∂ is a parameter term that can be changed arbitrarily and l is the distance value between two semantic elements.

In addition, when two semantic elements can be combined with a certain child node, these two distances can be expressed as 1 and there is no correlation between l and 1 in this case. In the case where l is greater than 3, theoretically the distance between these two semantic elements is relatively far and the strength of the relationship between them is 0.

Secondly, the strength of the relationship between words is interpreted according to the calculation results of Eq. (9). A, B The minimum relationship between the two words can be represented by the minimum set of relationships. Therefore, when analyzing the strength of the relationship between the words, the strength of the relationship between the words can be determined by analyzing the degree of common concepts, and the relationship expression is: (10) $D e g r e e (A, B) = \sum C (A_{i}, B_{j})$

Where $C (A_{i}, B_{j})$ is the dataset from 0 to 1, and C is the relationship strength value between words.

Select two learning resource words with hierarchical relationship in training set W_in to form a group. Use the training resources to place a window size in training set W_in to find two words and calculate the distance between element $I_{1} = {W_{1}, W_{2}}$ and the learning resource. Extract the relationship between the words through the training samples and select the elements obtained from the features I₂. Filter I₁ using I₂ to obtain I₃, $I_{3} = {W_{1}, W_{2}, W_{3}, W_{4}}$ . Calculate the strength of the relationship word pairs I₃ using Eq. (5) to perform filtering activities and classify the stored names according to the different strengths of the relationship to achieve the classification of different types of learning resources. 3)

Language Translation Corpus Resource Sharing [27]

The article obtained semantic concepts from the e-learning repository to acquire knowledge points related to the teaching program, which can ensure the authenticity and validity of the knowledge points. The similar features of teaching resources can be identified to the maximum extent, and their semantic similarity is calculated to implement the mode of nearest distance selection to quickly realize the exchange of resources. It should be noted that the calculation of resource similarity needs to be performed under a fixed threshold. The calculated resource similarity for C₁ and C₂ is: (11) $s i m (C_{1}, C_{2}) = D e g r e e (A, B) \frac{| C o u n t [U (C_{1}) \cap U (C_{2})] |}{| C o u n t [U (C_{1}) \cup U (C_{2})] |}$

Where: C₁ and C₂ are two similar teaching resources, respectively.

$C o u n t [U (C_{1}) \cap U (C_{2})]$ and $C o u n t [U (C_{1}) \cup U (C_{2})]$ are the simplified resource similarity and difference degrees using 1 + X certificates, respectively. By calculating the computational resource similarity for C₁ and C₂, the circulation between resources can be improved and the sharing of teaching resources can be maximized.

2.4

Resource-sharing capacity realization

2.4.1

Installing KVM for server-side virtualization

The use of KVM-based server virtualization technology: in the virtualization of the server side to achieve the functional modules include kernel-based virtualization management platform KVM, used to emulate the client operating system and its hardware platform on the open-source virtualization software QEMU, support for the QEMU device driver libvirt virtualization environment management suite and so on.

2.4.2

Installation and implementation of QEMU-KVM virtual machine

The implementation of this paper is based on the QEMU-KVM architecture, in which there are two main threads to implement two virtualization jobs respectively. The QEMU implements the I/O emulation and manages other things related to timers, including interrupts, displays; while the other part of the virtualization job is managed by the KVM, including the virtual CPU, memory, and so on. Therefore, QEMU-KVM is a user space management tool that can control KVM. The communication between QEMU and KVM is handled by /proc/kvm, /proc/vm and /proc/vcpu devices.

2.4.3

Installing the libvirt Virtualization Management Suite

Libvirt is a set of common APIs that provide functions in different languages to connect to the KVM to control the virtual machine with simple commands. In addition to the public API, Libvirt also provides a GUI-based tool called Virt-Manager for managing virtual machines. Libvirt supports a number of languages, such as Python, C, and Java, etc. In this paper, we utilize the Java API library to manage the guest operating system.

2.4.4

Application of the RDP protocol

Realization of RDP-based applications have three components, namely, the user interface for the transmission of communication information transmission protocol, the user’s client in the personal computer, and in the remote terminal server. The process of forming the data header in the RDP protocol is to initialize the parameters, in accordance with the rules of the network transmission to generate the standard header, the processing of the generation of the RDP-specific header; header sent for transmission that Give the user’s operation and feedback on the client.

2.4.5

Web application implementation

Perform the installation and configuration of Tomcat and MySQL, determine the implementation process of Struts2, through the Model layer to handle business logic, View layer to realize the interactive interface.

3

Experiments on Recommended Performance of English Translation Corpus Resource Cloud Platform

In order to validate the effectiveness of personalized recommendation of English translation corpus resources in the English translation corpus resource cloud platform proposed in this research paper, comparative experiments will be selected for validation in this chapter. User-based collaborative filtering recommendation algorithm (i.e., UserCF), latent trajectory modeling recommendation algorithm (i.e., LTM), and Markov chain prediction algorithm (i.e., MarkovChain) are introduced to compare them with the algorithm of this paper. The dataset used for this experiment is the International Knowledge Discovery and Data Mining Competition (i.e., KDD-Cup2010). The performance comparison results of this paper’s algorithm with UserCF, LTM, and Markov Chain prediction algorithms are specifically shown in Figure 1.

In the accuracy index comparison of this paper’s algorithm growth is more stable, the value is maintained between 0.2 and 0.25, the growth rate is not more than 0.1, while UserCF algorithm shows a decreasing trend, in the training set proportion of 2 when the value is as high as 52.6 or so, but for the 262,144 when the value is already lower than 20%. MarkovChain algorithm shows a trend of first decrease and then increase, the overall increase is more than 0.2.

In the recall comparison, the gradual increase in the proportion of the training set makes UserCF and MarkovChain algorithms show significant changes, with the former showing increasing recall and the latter showing fluctuating decrease. While this paper’s algorithm and LTM algorithm show a stable trend, the value is maintained between 0.04 and 0.05.

In the comparison of item coverage, MarkovChain algorithm shows a rapid decrease in item coverage, UserCF algorithm and LTM algorithm show an obvious trend of decreasing and then increasing and increasing and then decreasing, while this paper’s algorithm is obviously more stable, with the value maintained between 150 and 200.

In the comparison of AD results, the overall stability of this paper’s algorithm is the strongest, and the fluctuation ups and downs change the least. Comprehensively, the personalized learning resource recommendation algorithm proposed by the study in different sizes of training materials has high stability and the best recommendation performance.

In order to verify the recommendation effect of the English translation corpus resource cloud platform under the resource recommendation algorithm of this paper, the study has analyzed its practical application, conducted 100 and 500 experiments, and the results are shown in Table 1. As can be seen from the table, this paper’s algorithm in the actual operation of the cloud platform, 100 test cases not only the actual recommended response time of 4.5s, while the recommended success rate is as high as 99.6%. In 500 tests, the response time is 21.8s, the response time increases, but the success rate of recommendation is still as high as 99.5%. This shows the effectiveness of the personalized resource recommendation algorithm of this paper in the cloud platform system.

Table 1.

Experimental results

100 experimental results
Recommended response time	4.5s	Concurrency times	100
Throughput	23.6	Recommend the actual total number of responses	100
The actual average time of server request waiting	43.3ms	The average waiting time of user requests	4332.7ms
Recommended success rate	99.60%
500 experimental results
Recommended response time	21.8s	Concurrency times	500
Throughput	22.7	Recommend the actual total number of responses	500
The actual average time of server request waiting	43.6ms	The average waiting time of user requests	21923.1ms
Recommended success rate	99.50%

4

Experiments on Sharing Performance of English Translation Corpus Resource Cloud Platform

In order to verify the effectiveness of the sharing function of the English Translation Corpus Resource Cloud Platform designed in this paper, this chapter applies the English Translation Corpus Resource Cloud Platform to the teaching of English translation in T colleges and universities.

4.1

Evaluation of the English Translation Corpus Resource Cloud Platform

Invite 100 teachers and 100 students from T-University to use and experience the cloud platform for 30 days, and then use the questionnaire to conduct a survey. 220 questionnaires were sent out, 215 questionnaires were returned, and 215 questionnaires were valid.

4.1.1

Evaluation of the effectiveness of resource sharing

Table 2 shows the evaluation of the effect of English translation corpus resource sharing by teachers and students. It can be seen from the table that the evaluation results of teachers and students before and after the application of the cloud platform have obvious changes, and the mean values of various indicators have increased compared with those before the experiment, among which the mean values of the four indicators of “understanding the sharing of teaching resources”, “understanding of teaching resources”, “characteristics of teaching resource sharing” and “improvement of learning interest” have increased by 0.42, 0.05, 0.12 and 0.33 respectively before the experiment (P<0.05). Obviously, the interest and awareness of teachers and students have increased significantly, and the students’ learning autonomy and interest in learning have also improved.

Table 2.

Evaluation of resource sharing effect

Index	Before experiment	After experiment	T	P
Understanding the sharing of teaching resources	2.22	2.64	-10.2	0.001
Understanding of Teaching Resources	2.58	2.63	-3.111	0.003
Characteristics of teaching resource sharing	3.16	3.28	2.595	0.02
Students ‘attention concentration	3.07	3.12	1.965	0.066
Increased interest in learning	2.46	2.79	3.196	0.004

4.1.2

Overall evaluation of the cloud platform

The evaluation of this paper’s English translation corpus resource cloud platform by teachers and students after application is shown in Table 3. As can be seen from the table, in terms of the independent utilization of English translation corpus teaching resources, the majority of the people who hold a positive attitude towards the cloud platform of this paper are positive about the improvement of the ability to utilize resources independently (73.02%) and the improvement of self-competence (85.12%), and basically no longer worry about the source of resources (73.95%). In terms of teamwork and communication, teachers and students mostly agree that “collaborative communication contributes to the cultivation of teamwork ability” (82.79%), and the number of teachers and students who like teamwork and communication (65.12%) and think that discussion contributes to the rational use of resources (58.14%) is relatively small, respectively, 98 and 87, but the number of teachers and students who agree is relatively small. 98 and 87, but the percentage still exceeds the 50% level, and the group with a positive attitude still dominates. In terms of information literacy, those who believe that the design of the cloud platform in this paper helps typing efficiency and learning knowledge (95.81%), provides resource processing and handling ability (90.23%) and knowledge expansion (72.56%) dominate. In general, after applying the English translation corpus resource cloud platform constructed in this paper, teachers and students held a positive and affirmative attitude towards the cloud platform as a whole, and the sharing of English translation corpus resources by teachers and students was satisfied to a greater extent.

Table 3.

Overall evaluation of cloud platform

-	Positive attitude		Negative attitude
-	Number	Proportion	Number	Proportion
Independent use of teaching resources
It helps to improve the ability of independent use of resources	157	73.02%	58	26.98%
Like independent learning to use resources	183	85.12%	32	14.88%
Worry about the source of resources	56	26.05%	159	73.95%
Team collaboration communication
Like teamwork communication	140	65.12%	75	34.88%
The discussion is helpful to the rational utilization of resources.	125	58.14%	90	41.86%
Collaborative communication is helpful to the cultivation of team cooperation ability.	178	82.79%	37	17.21%
Information literacy
It helps to improve typing efficiency and learn knowledge.	206	95.81%	9	4.19%
It helps to improve the ability of resource processing.	194	90.23%	21	9.77%
Contribute to the expansion of knowledge	156	72.56%	59	27.44%

4.2

Experimental Analysis of English Translation Teaching

In this section, we will apply the English Translation Corpus Resource Cloud Platform proposed in this paper to the teaching of English translation and explore the effect of its sharing function in the actual teaching of English translation. In this study, a total of 50 subjects were randomly selected from T colleges and universities and randomly divided into 2 groups, the experimental group and the control group, with 25 subjects in each group. The experimental group used the cloud platform of this study for online English translation learning, while the control group only used the traditional resource library system provided by the university for English translation learning. This experiment is planned to be a 3-week learning activity, and the feasibility and validity are verified by means of tests, questionnaires and face-to-face interviews.

4.2.1

English translation test scores

The English translation test scores of the students in the experimental group and the control group before and after the experiment are specifically shown in Table 4. Before the experiment, the average score of the experimental group was 77.79 and the average score of the control group was 77.25, and the difference between the two scores is still less than 1, which indicates that the two groups of experimental subjects have equal mastery of the basic knowledge of learning. From the point of view of the results after the experiment, the average performance of the two groups of students have been improved, the experimental group’s average pre-test scores were 78.81 post-test scores averaged 81.52, with an average increase of 3.7. The post-test scores of the control group were increased from 77.28 to 78.71 from the pre-test, with an increase of 1.53. It can be seen that the experimental group’s performance improvement is higher than the control group’s, and the experimental treatment can be preliminarily judged that the adopting blended learning mode for vocational education selective courses in the experimental group is effective. There is also a statistically significant difference between the two groups when independent samples t-test was conducted (p=0.026<0.05).

Table 4.

The results of English translation test

Group	N	Mean	Standard deviation	T	P
Experimental group	25	77.82	9.348	0.215	0.832
Control group	25	77.28	9.229	0.215	0.832
Experimental group	25	81.52	7.774	0.784	0.026
Control group	25	78.81	8.133	0.784	0.026

4.2.2

Psychology of English Translation Learning

The learning psychology of English translation mainly includes five indicators: learning style, learning interest, attitude motivation, mental load and mental effort. The learning psychology of the experimental group and the control group after the experiment is specifically shown in Table 5. It can be seen that the mean values of the experimental group in the indicators of learning styles, learning interest and attitude motivation are higher than those of the control group by 14.53, 16.99 and 6.33 respectively, and all of them show significant differences (P<0.05). Obviously, the students in the experimental class had a more obvious improvement in learning style, learning interest, and attitude motivation. As for the two negative psychological indicators of mental load and mental effort, the mean values of the experimental group are lower than those of the control group, and they all present significant differences in the mental load indicators (P=0.006<0.01). In this paper, the English translation corpus resource cloud platform and its sharing function can effectively reduce the students’ corpus resource pressure on English translation learning, and the tendency of mental effort caused by learning pressure is also alleviated.

Table 5.

The learning psychology of English translation

Index	Group	N	Mean	Standard deviation	T	P
Learning mode	Experimental group	25	34.72	4.441	12.834	0.001
Learning mode	Control group	25	20.19	3.348	12.834	0.001
Study interest	Experimental group	25	32.56	2.115	24.726	0.008
Study interest	Control group	25	15.57	2.598	24.726	0.008
Attitude motivation	Experimental group	25	20.57	3.246	8.955	0.002
Attitude motivation	Control group	25	14.24	1.354	8.955	0.002
Mental load	Experimental group	25	13.33	4.716	-3.552	0.006
Mental load	Control group	25	17.43	3.36	-3.552	0.006
Mental effort	Experimental group	25	7.69	2.738	-1.186	0.224
Mental effort	Control group	25	8.61	2.748	-1.186	0.224

4.2.3

Cloud Platform Usage Experience

This subsection investigates the experimental group’s students’ experience of using the cloud platform in terms of practicality, operational simplicity, and the feeling of using it, and the results of the survey are shown in Figure 2. The number of items in the three aspects of practicality, ease of operation, and the feeling of use are 6, 7, and 5, respectively. This survey adopts a Likert 5-point scale to measure the students’ satisfaction with the learning method of this experiment.1 represents very dissatisfied, 2 represents relatively dissatisfied, 3 represents average, 4 represents relatively satisfied, and 5 represents very satisfied. It can be seen from the figure that the students in the experimental group are more satisfied and accept the practicality of the cloud platform in this paper, and the mean value of each question is more than 4 (27.88>4*6), that is, the frequency distribution graph is on the right side of the “more compatible”. In terms of ease of operation, the mean value of 32.14 is greater than 28 (32.14>4*7), and the overall distribution is on the right side of “more compatible”. The mean value of the scale score in the aspect of the feeling of use is 23.18 > 20 (23.18 > 4*5), which indicates that most of the students in the experimental group are satisfied or very satisfied with the experience of using the cloud platform. Therefore, it can be concluded that most of the teachers and students are satisfied with the acceptance of the English Translation Corpus Resource Cloud Platform used in this experiment, indicating that the Cloud Platform in this paper is scientific and reliable.

5

Conclusion

In this paper, cloud computing information technology is used as the theoretical support to build a cloud platform of English translation corpus resources to realize the sharing of English translation corpus resources in colleges and universities, and to provide assistance for the long-term development of English translation teaching. The core functions of the English translation corpus resource cloud platform mainly include the recommendation and sharing of English translation corpus resources, for which performance experiments are conducted to test the effectiveness of its functions.

In the recommendation performance experiments, comparative experiments are used for verification. The accuracy of this paper’s resource recommendation algorithm grows steadily between 0.2 and 0.25, the recall rate maintains a stable trend between 0.04 and 0.05, and in the comparison of the amount of item coverage and AD results, this paper’s algorithm also remains stable, with the smallest fluctuations and changes, and it has a better stability compared to the UserCF, LTM, and Markov Chain Prediction algorithms as a whole. Further analysis of the practical application of this paper’s resource recommendation algorithm through 100 and 500 experiments, this paper’s algorithm in the 100 and 500 experiments in the recommendation success rate as high as 99.6%, 99.5%, the corresponding response time of 4.5s, 21.8s. 500 experiments in the response time has increased, but still maintain a high recommendation success rate, this paper’s recommendation algorithm The overall recommendation algorithm in this paper still shows good performance.

In the sharing performance experiment, 100 teachers and 100 students from T University were invited to use and experience the cloud platform of this paper and collect evaluation feedback questionnaire data. In terms of the experience of resource sharing, the mean values of the four indicators of “understanding of teaching resource sharing”, “understanding of teaching resources”, “characteristics of teaching resource sharing” and “improvement of learning interest” of teachers and students all increased, which were increased by 0.42, 0.05, 0.12 and 0.33 respectively compared with those before the experiment (P<0.05). In the overall evaluation of the cloud platform, teachers and students who have a positive attitude towards improving the ability to utilize resources independently (73.02%) and improving self-competence (85.12%), recognize the view that “collaborative communication helps to cultivate the ability of teamwork” (82.79%), and like collaborative communication in teams (65.12%) and believe that discussion helps the rational use of resources (58.14%) are in the mainstream, both exceeding the 50% level. The proportion of teachers and students who think that discussion helps the rational use of resources (58.14%) is dominant, both exceeding the 50% level. In terms of information literacy-related feedback, those who thought that the design of the cloud platform in this paper helped typing efficiency and learning knowledge (95.81%), provided resource processing and handling ability (90.23%), and knowledge expansion (72.56%) were in the majority. Overall, the teachers and students who participated in the experiential testing had an overall positive and affirmative attitude towards the cloud platform.

Applying this paper’s English Translation Corpus Resource Cloud Platform to the actual teaching of English translation, the posttest score of the experimental group using the Cloud Platform improved by 3.7 points compared with the pre-test, reaching 81.37 points, which is higher than that of the control class by 2.76 points, showing a significant difference (P=0.026<0.05). The mean values of learning style, learning interest, and attitude motivation indicators of learning psychology in the experimental group were higher than those of the control group by 14.53, 16.99, and 6.33, respectively, presenting a significant difference (P<0.05). On the two negative psychological indicators of learning, mental load and mental effort, the experimental group presented a significant difference only in the mental load indicator (P=0.006<0.01), but the mean values of both indicators were lower than those of the control group. In addition, the experimental group is in a state of comparative satisfaction or great satisfaction with the cloud platform’s This paper’s cloud platform’s practicality, simplicity, and feeling of use in terms of measurement scores of 27.88, 32.14, and 23.18, respectively.

Language:: English

Publication timeframe:: 1 times per year
Journal Subjects:: Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other

Journal RSS Feed

Research on Resource Sharing Methods of English Translation Corpus in Colleges and Universities under the Background of Informatization

Wei Huang

Haiyan Zhang

Published Online: Sep 26, 2025

Received: Jan 31, 2025

Accepted: May 09, 2025

DOI: https://doi.org/10.2478/amns-2025-1066

KeywordsResource sharing, Cloud platform, Recommendation algorithm, Immuno-evolutionary algorithm

© 2025 Wei Huang et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
Resource sharing, Cloud platform, Recommendation algorithm, Immuno-evolutionary algorithm