A bilingual translation and life circle map recommendation co-creation system for international students

Abstract

The application discloses a bilingual translation and life circle map recommendation co-creation system for international students, comprising: a corpus management module for storing and managing hierarchical bilingual life corpus; a database module for storing map data and point of interest information related to international students; a core service engine module in communication connection with the corpus management module and the database module respectively; a user co-creation interaction module for receiving new corpus submitted by users, correction and supplement of existing corpus, and multilingual evaluation, pictures, strategies and scenario-based corpus use instances submitted for points of interest; an audit and incentive module for auditing the submitted content and synchronizing to the corpus management module and the database module, and granting rewards to users who contribute effective content. The application can provide accurate scenario-based translation and personalized life circle recommendation for international students, and realize self-improvement and ecological development of system data by encouraging users to contribute content.

Description

Technical Field

[0001] This invention relates to the field of Internet application technology, and in particular to a bilingual translation and community map recommendation co-creation system for international students. Background Technology

[0002] With the deepening development of globalized education, the number of international students continues to expand, and adapting to life and communicating in foreign countries has become a common core issue. Currently, language and living services for international students are often fragmented, making it difficult to form a coordinated support system.

[0003] In terms of language translation, existing translation tools mostly focus on general text translation scenarios, lacking professional and scenario-based bilingual corpora for high-frequency life scenarios of international students (such as campus registration, medical consultation, daily shopping, transportation, etc.). In particular, the coverage of local slang and common colloquialisms is insufficient, resulting in poor practicality of translation results. At the same time, the corpus content is mostly provided unilaterally by the platform, and users cannot participate in the supplementation, correction and scenario adaptation of the corpus, making it difficult to meet the personalized language needs of international students in different regions.

[0004] In terms of lifestyle services and location recommendations, the general map recommendation tool does not fully meet the specific needs of international students. The accuracy of the recommended points of interest (such as service facilities around the campus and student-friendly places) is insufficient, and it is not linked to the language translation function. After arriving at the recommended location, international students still face contextual language communication barriers. In addition, the information updates of points of interest rely on platform maintenance, and users cannot add co-created content such as location reviews and practical guides in real time, resulting in insufficient timeliness and richness of location information.

[0005] In existing technologies, language translation and location recommendation functions are independent of each other, failing to provide an integrated solution for international students. At the same time, the lack of an effective user co-creation mechanism and incentive system makes it difficult to gather collective wisdom to continuously optimize the quality of the corpus and location information, resulting in limited adaptability and iteration capabilities of the service. Summary of the Invention

[0006] The purpose of this invention is to provide a bilingual translation and community map recommendation co-creation system for international students. This invention can provide international students with accurate contextualized translations and personalized community recommendations, while simultaneously incentivizing user content contributions to achieve self-improvement and ecological development of the system's data.

[0007] The technical solution of this invention: A bilingual translation and community map recommendation co-creation system for international students, comprising: The corpus management module is used to store and manage a hierarchical bilingual everyday language corpus. Each corpus record in the bilingual everyday language corpus is associated with at least one everyday scene tag, original text, target translation, and user contribution identifier. The database module is used to store map data and points of interest information related to international students; The core service engine module communicates with the corpus management module and the database module respectively, and is used to respond to user requests, perform bilingual translation queries, intelligent recommendation of points of interest based on geographical location and scene, and integrate and display the corpus translation results with the relevant information of the recommended locations; The user co-creation interaction module is used to receive new corpora submitted by users, corrections and supplements to existing corpora, as well as multilingual evaluations, pictures, guides and scenario-based corpus usage examples submitted for points of interest. The review and incentive module is used to review the content submitted through the co-creation interface and synchronize it to the corpus management module and the database module, while also awarding rewards to users who contribute valid content.

[0008] The aforementioned bilingual translation and lifestyle map recommendation co-creation system for international students employs a multi-level classification structure for its bilingual lifestyle corpus. The first-level classification includes dining, shopping, transportation, accommodation, academics, healthcare, and social entertainment. Each first-level classification has further subdivided second-level scenario classifications. Each corpus record is dynamically linked to its query and usage frequency and user feedback ratings. The bilingual lifestyle corpus incorporates a corpus priority algorithm to accurately select the most suitable and practical corpus for the user's current scenario. The calculation formula for the corpus priority algorithm is as follows: ; In the formula: Prioritize the corpus. For corpus queries, use frequency normalization values. The average rating based on user feedback. The degree of label matching between the corpus and the current scene. , and These are all corresponding weighting coefficients.

[0009] The aforementioned bilingual translation and community map recommendation co-creation system for international students includes, in its database module, locations within the campus community, popular city attractions, and places frequently visited by young students, along with associated geographic coordinates, type tags, and multilingual descriptions.

[0010] The aforementioned bilingual translation and community map recommendation co-creation system for international students includes the following core service engine module: The contextualized translation unit is configured to retrieve and present the Top-N high-priority corpus translations from the corpus management module based on the user's input or selection of the current life scenario, through corpus priority calculation. The map recommendation unit is configured to generate a geographical area of ​​an "X-minute living circle" based on the user's current location or a specified location, combined with time parameters. It then filters and ranks points of interest using a point-of-interest (POI) recommendation scoring algorithm. The formula for this POI recommendation algorithm is: ; In the formula, For points of interest With user location Distance adaptability To reach the point of interest Is the time taken ≤ X minutes? For points of interest With target scene tags semantic similarity, For points of interest Normalized user traffic value , , and These are all corresponding weighting coefficients; The integrated display unit is configured to simultaneously call and embed high-frequency or high-rated bilingual corpus from the corpus management module that matches the type or scenario of the interest point when displaying details of recommended interest points to the user.

[0011] The aforementioned bilingual translation and social circle map recommendation co-creation system for international students further configures the map recommendation unit to: use user-selected scene tags or input corpus query keywords as semantic filtering conditions for recommended points of interest; calculate the relevance degree using a corpus point of interest relevance algorithm; and filter points with a relevance degree ≥ a threshold. The system uses points of interest to achieve a correlation recommendation from translation needs to location recommendations; the calculation formula for the corpus interest point correlation degree algorithm is as follows: ; In the formula, For corpus With points of interest The number of common tags, For corpus With points of interest The total number of tags.

[0012] The aforementioned bilingual translation and social circle map recommendation co-creation system for international students includes a bilingual tag system in its database module. This system comprises both system-defined and user-co-created tags, including official standard tags and user-generated colloquial and contextualized tags. A semantic similarity algorithm is used to calculate tag similarity, achieving semantic association with the corpus records in the corpus management module. The formula for the semantic similarity algorithm is as follows: ; In the formula, , For tags , Word vectors, For dot product operation, Let be the vector magnitude.

[0013] The aforementioned bilingual translation and community map recommendation co-creation system for international students also includes a task guidance mechanism in its user co-creation interaction module. This mechanism calculates task priorities using a co-creation demand urgency algorithm, prioritizes high-priority co-creation tasks for users, and guides users to conduct on-site information collection and verification. The formula for calculating the co-creation demand urgency algorithm is as follows: ; In the formula, The degree of missing target content, Based on the user demand for this type of content, and All are weighting coefficients.

[0014] The aforementioned bilingual translation and community map recommendation co-creation system for international students quantifies the review results of the review and incentive module through a content quality scoring algorithm. ; In the formula, For content validity, To ensure the uniqueness of the content, The relevance of content to the corresponding corpus or points of interest. , and These are all corresponding weighting coefficients.

[0015] The aforementioned bilingual translation and community map recommendation co-creation system for international students includes a review and incentive module where rewards are linked to the quality level, frequency of use, and recognition of user-contributed content. These rewards include points, reputation levels, or rights.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention integrates scenario-based bilingual translation with interest-based recommendations in daily life circles, providing international students with an integrated service of language communication and location navigation. It not only solves the translation pain points of high-frequency daily life scenarios, but also accurately matches international students' exclusive interest points, greatly reducing the cost of adapting to life abroad.

[0017] 2. This invention constructs a hierarchical bilingual everyday language corpus, covering core scenarios such as catering, medical care, and academia, as well as colloquial expressions. It dynamically presents highly adaptable corpora using a priority algorithm, while also supporting user co-creation and supplementation, so that the corpus continuously meets the personalized and localized needs of different international students, thereby improving the practicality of translation.

[0018] 3. This invention achieves accurate recommendations from multiple dimensions, including location, scene, and translation needs, by limiting the geographical scope of the "X-minute living circle," calculating semantic similarity, and associating corpus interest points with algorithms. At the same time, it incorporates user-co-created reviews and guides to ensure the timeliness and richness of interest point information.

[0019] 4. This invention relies on a user co-creation interaction module and an audit incentive system to gather collective wisdom to continuously optimize the corpus and interest point database, so that the system's adaptability and service capabilities are continuously improved as users use it.

[0020] 5. This invention embeds matching high-frequency / high-rated bilingual corpora into the details of points of interest, enabling simultaneous display of location recommendations and contextualized translations, avoiding language communication barriers faced by international students upon arrival at their destinations, and improving the continuity and ease of use of the service. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a structural diagram of a bilingual everyday language corpus; Figure 3 This is a structural diagram of the core service engine module. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the present invention.

[0023] Example: A co-creation system for bilingual translation and community map recommendations for international students, such as... Figure 1 As shown, it includes: The corpus management module is used to store and manage a hierarchical bilingual everyday language corpus. Each corpus record in the bilingual everyday language corpus is associated with at least one everyday scene tag, original text, target translation, and user contribution identifier. The database module is used to store map data and points of interest information related to international students; The core service engine module communicates with the corpus management module and the database module respectively, and is used to respond to user requests, perform bilingual translation queries, intelligent recommendation of points of interest based on geographical location and scene, and integrate and display the corpus translation results with the relevant information of the recommended locations; The user co-creation interaction module is used to receive new corpora submitted by users, corrections and supplements to existing corpora, as well as multilingual evaluations, pictures, guides and scenario-based corpus usage examples submitted for points of interest. The review and incentive module is used to review the content submitted through the co-creation interface and synchronize it to the corpus management module and the database module, while also awarding rewards to users who contribute valid content.

[0024] In this embodiment, as Figure 2 As shown, the bilingual everyday language corpus adopts a three-level hierarchical structure: primary scene classification, secondary scene subdivision, and corpus recording. The primary categories are fixed as seven major categories: catering, shopping, transportation, accommodation, academic, medical, and social entertainment, covering the core life scenarios of international students; The secondary scenario categories are further subdivided according to actual needs. For example, under the catering category, there are categories such as "restaurant ordering", "takeout ordering", and "ingredient procurement", while under the transportation category, there are categories such as "bus ride", "subway ticket purchase", and "ride-hailing". Each subdivided scenario is marked with a unique scenario code (e.g., catering-ordering: C001). Each corpus record uses a structured data format, with fields including: Corpus ID (unique identifier, format LC-YYYYMMDD-XXXX), Source Text (source language text, supports multilingual input), Target Translation (target language text, defaults to Chinese and English bilingual, expandable to multilingual), Scene Tags (associated with one or more secondary scene codes), User Contribution Identifier (contributing user ID and contribution timestamp), Query Frequency (cumulative number of queries), User Feedback Rating (average of valid ratings in the past 30 days, maximum score of 5), and Review Status (approved / pending review / not approved).

[0025] The bilingual everyday language corpus uses a MySQL database to store structured corpus fields and combines it with Elasticsearch to achieve efficient retrieval of corpus text, supporting fuzzy queries and accurate matching of scene tags. At the same time, a corpus update mechanism is established to periodically synchronize new corpus and corrections that have passed review, cache frequently queried corpus to improve retrieval response speed (target response time ≤300ms), realize corpus version management, retain historical correction records, and support rollback function for accidental operations.

[0026] The bilingual everyday language corpus includes a corpus priority algorithm to accurately select the most suitable and practical language data for the user's current situation. Specifically, this algorithm includes: Data preprocessing: The frequency of query usage in the corpus is normalized, and the original number of queries is mapped to the [0,1] interval.

[0027] Parameter values: weighting coefficients (Query frequency weight) (Scoring weight) (Tag matching weights) are set to 0.3, 0.4, and 0.3 by default, and can be dynamically adjusted based on system operation data; Tag matching degree calculation: If the associated tags in the corpus are completely consistent with the tags in the current scene, =1; if there is a partial match, then M = number of common tags / total number of tags in the corpus; if there is no match, then... =0; Priority sorting: by The results are sorted in descending order, and the top-N (default N=5, user-defined 1-10) corpora are taken as the recommendation results.

[0028] In this embodiment, the database module stores basic electronic map data, including road networks, administrative divisions, and geographic coordinate systems (using the WGS84 coordinate system). It updates road and traffic status data in real time by connecting to the Gaode Map / Baidu Map API. Points of Interest (POI) information is stored in a structured data table. Core fields include: POI ID (unique identifier, format: POI-YYYYMMDD-XXXX), name (multilingual name, supporting both Chinese and English), geographic coordinates (longitude × latitude, accurate to 6 decimal places), type tags (associated with official system tags and user-created tags, separated by commas), address information (detailed address and postal code), contact information (telephone and website link, optional), business hours, multilingual description (default Chinese and English, including venue features, student-friendly services, etc.), user visits (cumulative number of visits), average rating (average user rating, maximum 5 points), and a list of associated corpus IDs (a set of corpus IDs with a matching degree ≥ 0.8).

[0029] The bilingual tagging system is constructed as follows: Standard labels: set according to the primary scene category, such as "Chinese food", "Western food" and "fast food" for catering, and "library", "laboratory" and "teaching building" for academics. The label code is associated with the scene category. User-created tags: Users can submit colloquial and contextual tags (such as "discounts for international students", "Chinese services", "near subway station"), which will be included in the tag system after being reviewed and approved, and managed with the same coding as the official tags; Tag semantic association: Tag word vectors are trained using the Word2Vec model, mapping official tags and co-created tags to the same semantic space for semantic similarity calculation.

[0030] The database module automatically synchronizes the basic map data with third-party API updates every day at midnight; Points of Interest (POI) information supports both manual updates by system administrators and user-co-created updates, and user-submitted POI additions / corrections are synchronized in real time after review; in addition, the database module establishes a data redundancy backup mechanism, with daily full backups and incremental backups, and backup data is retained for 90 days.

[0031] In this embodiment, as Figure 3 As shown, the core service engine module includes: The contextualized translation unit receives the original text or selected scene tags input by the user, triggering the corpus retrieval process. If the user inputs the original text, it first matches the corresponding scene tags through semantic analysis (using the TF-IDF algorithm to extract keywords and semantically match them with the scene tags), and then combines the corpus priority algorithm to filter high-priority corpus. The contextualized translation unit supports text input and voice input (connected to the speech recognition API to convert to text), and the output translation supports text display and voice reading (connected to the TTS speech synthesis API), and provides entry points for corpus collection and feedback rating functions; the contextualized translation unit adopts a concurrent processing mechanism, supports multiple users querying simultaneously, and the maximum concurrency of a single service node is ≥1000 QPS.

[0032] The map recommendation unit is configured to generate a "X-minute living circle" geographical range based on the user's current location or a specified location, combined with time parameters. It then filters and sorts points of interest using an interest point recommendation scoring algorithm, and displays the top-10 points in descending order.

[0033] The geographic range is based on the user's current location (GPS positioning or manually entered location coordinates), combined with the mode of transportation (walking by default, but users can choose public transportation, cycling, or driving), and calculates the geographic boundary that can be reached in X minutes (X=15 by default, but users can customize 5-60 minutes) through the map API, forming a circular or polygonal living circle range.

[0034] The formula for the point of interest recommendation algorithm is: ; In the formula, For points of interest With user location The distance fit is set to [0,1], i.e. , For user location With points of interest The straight-line distance The maximum distance can be reached in X minutes. To reach the point of interest If the time taken is ≤X minutes, then the value is 1; otherwise, the value is 0. For points of interest With target scene tags The semantic similarity can be calculated using the dot product of the tag word vectors, for example... ; , For tags , Word vectors, For dot product operation, The vector magnitude; For points of interest Normalized user traffic value , , and These are the corresponding weighting coefficients, which can take values ​​of 0.3, 0.4, 0.2 and 0.1 respectively.

[0035] The map recommendation unit is further configured to: use the scene tags selected by the user or the keywords entered from the corpus query as semantic filtering conditions for recommended points of interest; calculate the relevance using a corpus point of interest relevance algorithm; and filter points with a relevance ≥ a threshold. The system uses points of interest to achieve a correlation recommendation from translation needs to location recommendations; the calculation formula for the corpus interest point correlation degree algorithm is as follows: ; In the formula, For corpus With points of interest The number of common tags, For corpus With points of interest The total number of tags.

[0036] threshold Set to 0.5; this value is configurable and adjustable. When a user queries a corpus or selects a scene tag, the system automatically calculates the correlation between the corpus / tag and all points of interest, and then filters the results. ≥threshold The system incorporates points of interest into the recommendation list, achieving a seamless connection between translation needs and location recommendations.

[0037] The integrated display unit is configured to simultaneously call and embed high-frequency or high-rated bilingual corpora from the corpus management module that match the type or scenario of the recommended points of interest when showing users details of those points. The integrated display unit's interface uses a two-layer display mode: a map and details. The map layer displays the location markers and the surrounding area; clicking on a marker brings up the details layer. While displaying basic information about the points of interest, the details layer calls the corpus management module to retrieve high-frequency (top 20% of queries) or high-rated (≥4.5) corpora matching the point's type and scenario tags. These corpora are then sorted and displayed according to their relevance to the usage scenario; for example, for restaurant points of interest, "common phrases for ordering food" and "menu inquiry corpora" are embedded. This unit also allows users to quickly copy and read aloud corpora by clicking on them and provides a feedback entry for "corpus-point matching degree" to optimize the accuracy of corpus embedding.

[0038] In this embodiment, the core service engine module uses Python as the algorithm development language and combines the TensorFlow framework to implement models such as word vector training and semantic similarity calculation. The algorithm module is encapsulated as an independent service and communicates with other modules through a RESTful API, supporting algorithm version iteration and independent deployment. At the same time, an algorithm monitoring mechanism is established to statistically analyze the recommendation accuracy (the proportion of users clicking on recommended points of interest) and the corpus matching accuracy (the proportion of user feedback corpus matching the scene) in real time. When the accuracy is lower than the threshold, an alarm is triggered, and the algorithm parameters are manually adjusted.

[0039] In this embodiment, the user co-creation interaction module's co-creation function entry points include: corpus contribution, interest point co-creation, and task guidance. Corpus contribution provides entry points for "Add Corpus" and "Revise Corpus." Adding corpus requires filling in the original text, translation, and associated scene tags; revising corpus requires selecting the target corpus ID and filling in the revision content. Interest point co-creation provides entry points for "Add Interest Point," "Evaluate Interest Point," and "Supplementary Guide." Adding an interest point requires filling in basic information such as name, coordinates, address, and type tags; evaluating interest points supports rating, multilingual text evaluation, and image upload; supplementary guides support scenario-based usage suggestions (such as "Library Study Guide" and "Hospital Visit Procedure"). Task guidance uses a homepage display of high-priority co-creation tasks (filtered based on a co-creation demand urgency algorithm), such as "Information on interest points in a certain area is missing" or "Corps in a certain scene is insufficient," guiding users to collect information on-site. Task details indicate task requirements and reward points. The task guidance mechanism calculates task priority using a co-creation demand urgency algorithm and prioritizes high-priority tasks for users. The co-creation task has an urgency level greater than 0.7, and users are guided to conduct on-site information collection and verification; the calculation formula for the co-creation demand urgency algorithm is as follows: ; In the formula, The degree of missing target content is calculated as follows: (corpus missing is calculated as "the amount of corpus in this scene / the amount of standard corpus"; interest point missing is calculated as "the density of interest points in this area / the standard density"). =1−actual value / standard value), with a value range of [0,1]. This represents the user demand popularity for this type of content (normalized value of the number of queries for this type of content in the past 30 days), with a value range of [0,1]. and These are all weighting coefficients, with values ​​of 0.6 and 0.4 respectively.

[0040] In this embodiment, the review and incentive module includes automatic preliminary review and manual secondary review. The automatic preliminary review uses a rule engine and content risk control model to initially screen co-created content, such as detecting whether the text contains prohibited words, whether the coordinates of points of interest are within a reasonable range, and whether the translated text matches the semantics of the original text (by comparing semantic similarity through machine translation). The manual secondary review involves assigning content that passes the automatic preliminary review to reviewer accounts. Reviewers then review the validity, uniqueness, and relevance of the content according to standards, and the review results are marked as "Passed" or "Rejected (with reasons for rejection)." The review timeframe for the review and incentive module is ≤24 hours for corpus-type content and ≤48 hours for point-of-interest and strategy-type content.

[0041] Furthermore, the review results of the review and incentive module are quantified using a content quality scoring algorithm: ; In the formula, For content validity, a value of 1 is assigned if the content meets the scenario requirements and the information is accurate; otherwise, a value of 0 is assigned. For content uniqueness, a value of 1 is assigned if the overlap with existing content is ≤30%, otherwise a value of 0 is assigned. The relevance of the content to the corresponding corpus or point of interest is denoted by the scene matching degree with the corresponding corpus / point of interest ≥ 0.7, which is 1 if it is ≥ 0, otherwise it is 0. , and These are the corresponding weighting coefficients, with values ​​of 0.5, 0.3, and 0.2 respectively.

[0042] When content quality rating Content with a score of ≥0.8 is considered high-quality, while content with a score of 0.5 or ≤0.8 is considered excellent. <0.8 is considered acceptable. <0.5 indicates an unacceptable item.

[0043] The incentive system is implemented as follows: Points Rewards: High-quality content earns 10-20 points, while qualified content earns 5-10 points. Points can be redeemed for platform benefits (such as ad-free services and advanced translation features) or physical gifts. Reputation Level: User reputation levels are divided into 1-5. Users who contribute ≥50 pieces of high-quality content will be upgraded to Level 2, ≥100 pieces will be upgraded to Level 3, and so on. Higher-level users can enjoy privileges such as expedited review and exclusive tasks. Points Redemption: Establish a points redemption mall, regularly update redeemable items and points redemption ratios, and support users to redeem online and track logistics delivery.

[0044] Rewards Distribution: Points will be distributed in real time after approval. Reputation level is updated on the 1st of each month. Users can view their contribution records, point balance, and reputation level in their personal center.

[0045] In summary, this invention can provide international students with accurate contextualized translation and personalized lifestyle recommendations, while also enabling the system's data to self-improve and develop ecologically by incentivizing users to contribute content.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the invention without departing from the spirit and scope of the present invention.

Claims

1. A bilingual translation and community map recommendation co-creation system for international students, characterized in that: include: The corpus management module is used to store and manage a hierarchical bilingual everyday language corpus. Each corpus record in the bilingual everyday language corpus is associated with at least one everyday scene tag, original text, target translation, and user contribution identifier. The database module is used to store map data and points of interest information related to international students; The core service engine module communicates with the corpus management module and the database module respectively, and is used to respond to user requests, perform bilingual translation queries, intelligent recommendation of points of interest based on geographical location and scene, and integrate and display the corpus translation results with the relevant information of the recommended locations; The user co-creation interaction module is used to receive new corpora submitted by users, corrections and supplements to existing corpora, as well as multilingual evaluations, pictures, guides and scenario-based corpus usage examples submitted for points of interest. The review and incentive module is used to review the content submitted through the co-creation interface and synchronize it to the corpus management module and the database module, while also awarding rewards to users who contribute valid content.

2. The bilingual translation and community map recommendation co-creation system for international students as described in claim 1, characterized in that: The bilingual everyday language corpus adopts a multi-level classification structure. The first-level classification includes dining, shopping, transportation, accommodation, academics, medical care, and social entertainment. Each first-level classification has a further subdivided second-level scenario classification. Each corpus record is dynamically associated with its query and usage frequency and user feedback rating. The bilingual everyday language corpus is equipped with a corpus priority algorithm to accurately select the corpus that is most suitable and most practical for the user's current scenario. The calculation formula of the corpus priority algorithm is as follows: ; In the formula: Prioritize the corpus. For corpus queries, use frequency normalization values. The average rating based on user feedback. The degree of label matching between the corpus and the current scene. , and These are all corresponding weighting coefficients.

3. The bilingual translation and community map recommendation co-creation system for international students as described in claim 1, characterized in that, In the database module, the point of interest information includes locations within the campus living area, popular urban attractions, and places frequently visited by young students, and is associated with geographic coordinates, type tags, and multilingual descriptions.

4. The bilingual translation and community map recommendation co-creation system for international students as described in claim 2, characterized in that, The core service engine module includes: The contextualized translation unit is configured to retrieve and present the Top-N high-priority corpus translations from the corpus management module based on the user's input or selection of the current life scenario, through corpus priority calculation. The map recommendation unit is configured to generate a geographical area of ​​an "X-minute living circle" based on the user's current location or a specified location, combined with time parameters. It then filters and ranks points of interest using a point-of-interest (POI) recommendation scoring algorithm. The formula for this POI recommendation algorithm is: ； In the formula, For points of interest With user location Distance adaptability To reach the point of interest Is the time taken ≤ X minutes? For points of interest With target scene tags semantic similarity, For points of interest Normalized user traffic value , , and These are all corresponding weighting coefficients; The integrated display unit is configured to simultaneously call and embed high-frequency or high-rated bilingual corpus from the corpus management module that matches the type or scenario of the interest point when displaying details of recommended interest points to the user.

5. The bilingual translation and community map recommendation co-creation system for international students according to claim 4, characterized in that, The map recommendation unit is further configured to: use the scene tags selected by the user or the keywords entered from the corpus query as semantic filtering conditions for recommended points of interest; calculate the relevance using a corpus point of interest relevance algorithm; and filter points with a relevance ≥ a threshold. The system uses points of interest to achieve a correlation recommendation from translation needs to location recommendations; the calculation formula for the corpus interest point correlation degree algorithm is as follows: ； In the formula, For corpus With points of interest The number of common tags, For corpus With points of interest The total number of tags.

6. The bilingual translation and community map recommendation co-creation system for international students according to claim 4, characterized in that, The interest point information in the database module includes a bilingual tag system defined by the system and co-created by users. This bilingual tag system includes official standard tags and user-generated colloquial and contextualized tags. The tag similarity is calculated using a semantic similarity algorithm to achieve semantic association with the corpus records in the corpus management module. The formula for the semantic similarity algorithm is as follows: ； In the formula, , For tags , Word vectors, For dot product operation, Let be the vector magnitude.

7. The bilingual translation and community map recommendation co-creation system for international students as described in claim 1, characterized in that, The user co-creation interaction module is also equipped with a task guidance mechanism. This mechanism calculates task priorities using a co-creation demand urgency algorithm, prioritizes high-priority co-creation tasks for users, and guides users to conduct on-site information collection and verification. The calculation formula for the co-creation demand urgency algorithm is as follows: ； In the formula, The degree of missing content in the target context. Based on the user demand for this type of content, and All are weighting coefficients.

8. The bilingual translation and community map recommendation co-creation system for international students according to claim 1, characterized in that, The review results of the review and incentive module are quantified using a content quality scoring algorithm: ； In the formula, For content validity, To ensure the uniqueness of the content, The relevance of content to the corresponding corpus or points of interest. , and These are all corresponding weighting coefficients.

9. The bilingual translation and community map recommendation co-creation system for international students according to claim 1, characterized in that, The rewards in the review and incentive module are related to the quality level, frequency of use, and recognition of the user's contributed content. The rewards include the granting of points, reputation levels, or rights.

Images