Today should be amplified
Here is the competition clearance guide——
The 2nd "Mobile Cloud Cup" Computing Power Network Application Innovation Competition
University Track - Nine-day Application Sub-track
Interpretation of the super version of the preliminary competition questions
Catch it quickly!
1
Background
Internet TV, or OTT TV (Over the Top TV), is a technology that utilizes a broadband cable TV network and integrates various technologies such as the Internet, multimedia, and communication to provide a variety of interactive services to home Internet TV users. How to use multi-scenario data to build a unified model to meet the recommendation requirements of different scenarios, and optimize the click-through rate and playback time at the same time has become a key issue in the Internet TV scenario. In addition, when recommending Internet TV, the interactive interface displays the poster picture of the media asset, and after the user clicks on the media asset information details page, there will also be a text introduction of the media asset. These multi-modal information plays an important role in the recommendation in the Internet TV scene. How to integrate multimodal information into the recommendation system is also an important research content in the Internet TV recommendation scenario.
2
Introduction
There are two competitions in this track:
1) Question 1 : "Multi-objective and multi-scenario video recommendation challenge" is the preliminary competition topic. The core task is to use the multi-modal information of media resources and user history logs in multiple scenarios to recommend media resources for users while taking into account the click-through rate. with playback duration indicators.
A. Dataset Introduction
The contest organizer will provide various modal feature data such as pictures of media assets, text introductions, and text labels, as well as user behavior data of 30 days in history. The behavior data includes 6 columns, user id, media asset id, scene id, Behavior timestamp, click or not, and whether the playback is complete. In addition, the contest organizer will also provide test samples, which include 5 columns, sample id, user id, media asset id, scene id, and behavior timestamp. Participants need to build a recommendation model based on user history and multi-modal feature data of media assets, and score each sample on the test set. The evaluation system will evaluate the contestants' recommendation results based on the user's actual interaction behavior records. The evaluation indicators mainly examine the recommended click accuracy (gAUC of clicked samples) and playback completion rate (gAUC of completed samples). The average calculation of multiple scene indicators is the final evaluation indicator.
B. The baseline environment depends on the following:
python==3.7.11
pandas==1.3.5
tensorflow==2.4.0
deepctr==0.9.1
numpy==1.19.5C. How to view and download data
https://ecloud.10086.cn/api/query/developer/s/Znq6za
2) Question 2 : Only full-time students (including vocational, undergraduate, and postgraduate students) of ordinary colleges and universities in Liaoning Province can register for the competition. For details of the competition, please refer to the Liaoning Provincial University Student Innovation and Entrepreneurship Management Sharing Platform (https://cxcy.upln .cn/match/details?comId=dffb08976a337cfd5826d3c7ecb152e1)
Entries must focus on current core fields and advanced technologies such as cloud computing, artificial intelligence, big data, and computing-network integration, and explore technical applications and innovative works with specific landing scenarios. Entries should have predictable social development and market value. The field can be applied to industry, agriculture, transportation, medical care, culture, education, finance, public safety, daily life, public welfare and other industries, such as smart city, smart industry, smart agriculture, smart transportation, smart campus, smart home , intelligent design, mobile application development, in-vehicle service, VR/AR and other cloud-intelligent fusion applications or ideas. (Note: Works that participated in the Mobile Cloud Cup Computing Power Network Application Innovation Competition held by China Mobile in 2022 are not considered valid works for this round).
Participants must have certain basic knowledge of IT, and can choose their own creative application scenarios of technology. Each participating team can design, produce, and finally submit original entries and proposals that can be displayed in visual applications based on their own interests, technical capabilities, and knowledge base. The forms of works include but are not limited to the use of various product capabilities opened by the mobile cloud, design and development of software products, smart terminal (such as smart phones, robots, software and hardware all-in-one machines, etc.) applications, etc.
3
Competition arrangement
The track is divided into five stages: registration, preliminaries, semi-finals, track finals, and national finals. 20 teams advance to the preliminary rounds, 10 teams advance to the semi-finals, and the top 3 teams in the track finals are recommended to be selected for the national finals. The team won the Track Excellence Award. The teams that have been promoted in each review stage should follow the unified arrangement of the competition to participate in the next round of competition review. If the team cannot participate due to personal reasons, it will be deemed as giving up the promotion quota for the next round, and the promotion quota (including bonus) will be postponed according to the ranking of this stage.
If you have any questions during the competition, you can contact the competition assistant for consultation.
official interpretation
Get twice the result with half the effort
I wish you all the best
Get good grades!