Maintaining stable and high-quality output, adept at seizing short-term trends, and utilizing diverse interactive techniques.
Author: murmurphy.eth
In the Web3 field, X (formerly Twitter) is undoubtedly the core battlefield for project operation and market expansion. For practitioners and investors, X is not only an excellent platform to enhance personal influence but also an opportunity to uncover new prospects. This article will unveil the recommendation algorithm and share practical operational strategies to help you quickly increase content exposure and interaction.
This article will analyze the algorithm logic from the perspective of X's official recommendation algorithm and share simple and actionable operational strategies to help you quickly break the ground.
Friendly reminder: Don’t want to see the details of the recommendation process? Jump directly to "Tweet Guidance Strategies and Rhythm Management" and the section on "Efficient Traffic Generation and Leverage Strategies" to get practical tips.
X (Twitter) Recommendation Algorithm Process
The core of Twitter's "For You" timeline recommendation system is to predict user interest in each tweet based on a large amount of user interaction data. The specific process is illustrated in the following diagram:
Step 1. Data Collection (Data) These data collectively form the "raw materials" of the recommendation system, providing a solid foundation for subsequent feature extraction and model training, ensuring that the algorithm can accurately capture user interests and behavioral characteristics.
Social Graph: The social graph collects the relationships and interactions between users, helping to build the user social network and providing a basis for identifying user social circles and active relationships in subsequent recommendations.
Tweet Engagement: Tracks various interaction behaviors of users with tweets, such as likes, comments, retweets, etc. This reflects the popularity of the content and the intensity of user interest, providing key data for the algorithm to assess tweet quality and relevance.
User Data: Collects information on users' personal preferences, usage habits, historical behaviors, etc., helping the model to more accurately identify and predict user interests and behavioral patterns.
Step 2. Feature Extraction (Features) After obtaining the raw data, the system uses a series of "tools" to deeply process this data, forming feature indicators for machine learning models.
GraphJet: A real-time graph engine used to analyze the bidirectional interactions between users and tweets.
RealGraph: Captures real social relationships and interaction patterns.
SimClusters: Uses clustering algorithms to group users or tweets with similar interests, discovering potential associations.
TwtNN: A deep learning model capable of extracting multidimensional features, thus more accurately capturing user interests.
TweepCred: Measures user credibility and influence on the platform, providing a reference for trust assessment.
Trust & Safety: Specifically responsible for detecting and filtering inappropriate or harmful content, ensuring the safety and compliance of recommended content.
Through these tools, the system can transform complex raw data into structured features, laying a solid foundation for subsequent algorithms to accurately determine which tweets are worth recommending.
Step 3. Candidate Source In the candidate generation phase, the system quickly filters potential content that matches user interests from a vast number of tweets, providing a data foundation for subsequent ranking. This phase is mainly achieved through four channels:
Search Index: Extracts tweets related to current trends through keywords or popular searches.
CR Mixer: Mixes, deduplicates, and filters the candidate list, outputting it to the subsequent core ranker.
UTEG: Establishes a relationship graph between users, tweets, and the keywords or topic entities within them, helping the algorithm understand the deeper connections between users and content.
FRS: Focuses on recommendations that can identify accounts you may be interested in but have not yet followed, introducing their tweets as candidates to increase account diversity and help you discover more quality content.
Overall, this phase ensures that the candidate list contains both current hot topics and reflects users' long-term interests through multidimensional and multi-channel filtering, laying a solid foundation for precise ranking.
Step 4. Ranking Engine (Heavy Ranker) In this phase, the system uses deep neural networks to evaluate each candidate tweet. It first calculates the predicted probabilities of user interactions (such as likes, comments, retweets, etc.), then multiplies these probabilities by preset weights and sums them up, ultimately obtaining a comprehensive score for each tweet. The higher the score, the more likely the tweet will appear on the timeline.
According to the default weights released by Twitter on GitHub on April 5, 2023, the weights and meanings of different interactions are roughly as follows:
By weighting various positive and negative interactions, the Heavy Ranker can quickly identify which content is most likely to be favored by users and which content should be reduced in recommendations.
Step 5: Rules and Filtering (Heuristics & Filtering) The ranked content will undergo a series of rule adjustments to ensure that the recommended content is both diverse and meets platform requirements. This process checks the overall popularity and social recognition of tweets while focusing on the diversity of authors to avoid excessive content from the same source. Additionally, if tweets contain violations, sensitive information, or excessive repetition, the system will ensure user experience and content safety by reducing visibility or filtering. This step acts as the "final checkpoint," downgrading or filtering out potentially duplicate, inappropriate, or unsuitable content.
Step 6: Mixing Output and Timeline Generation Finally, the system will add advertising content and new recommended accounts to the tweets that have been sorted and filtered, aiming to present you with a rich and balanced information timeline. The system will continuously adjust based on your new behavioral data to ensure that the content remains aligned with your interests.
In summary: Twitter filters and presents content that best matches user preferences through a process of data collection, feature extraction, candidate generation, ranking and filtering, and mixing output.
Tweet Guidance Strategies and Rhythm Management
To gain more exposure for tweets on the timelines of target users, it is essential to focus on the candidate generation and core ranking phases, ensuring that the content can be included in the candidate list and receives a high score during the ranking phase. Here are some effective tips:
1⃣ Basic Interaction: Promote Likes, Comments, and Retweets
Tweet replies have the highest weight score in the ranking engine, and extended interactions (e.g., the original author responding after a reply) can have a weight as high as 75.0. This bidirectional interaction is an extremely strong positive signal, indicating that the tweet not only attracted users but also stimulated further interaction from the author, significantly increasing exposure. Pose open-ended questions or controversial topics in tweets to attract reader comments. Also, don’t forget to actively respond to commenters' viewpoints and engage in in-depth discussions on users' questions or thoughts. This not only increases the number of replies but also makes readers feel valued, further enhancing account stickiness.
2⃣ Advanced Interaction: Guide Homepage Clicks and Extended Reading
Attract others to click on your homepage through a tweet, showing interest in browsing your other messages. Clicking on the author's homepage and generating other interactions (weight 12.0) indicates that users are not only interested in the content itself but also want to learn more about the author's other works.
You can set clear guidance in the tweet to encourage homepage clicks. Additionally, utilize pinned tweets or curated lists (Moments) to compile your best content, making it easier for homepage visitors to quickly access and generate more interactions.
3⃣ Stability and Outburst: Balance Long-term and Short-term Interactions
Twitter's "Long-term (50 days) + Short-term (3 days or even 30 minutes)" rolling aggregation features mean that the platform pays attention to your performance over a longer period while also tracking your recent or real-time dynamic performance. Specifically:
Long-term Performance: Regularly publish high-quality long-tail content (such as daily macro data analysis), continuously accumulating stable interactions and building brand weight.
Short-term Outburst: Seize hot topics and periods of fan activity, quickly output real-time dynamics, striving to achieve high interactions in a short time, enhancing content performance in short-term aggregated data.
Moreover, both long-term and short-term data are continuously updated, and the platform will "see" your new performance at any time. Therefore, it is recommended to regularly monitor interaction data and follower growth. If a decline in short-term interactions is detected, promptly adjust topics or posting times to avoid impacting long-term data performance.
4⃣ Group Interaction: Expand Social Graph and Stimulate Natural Discussions
By genuinely mentioning accounts, the platform's RealGraph will capture natural interactions between users, while the GraphJet will update your social graph data in real-time. This not only helps identify active users but also allows more users to see the connections between you and your partners, thus gaining additional exposure. Interactions in the comments section are also very valuable, such as liking/replying to comments and staying for more than 2 minutes, with weights of 11.0 and 10.0, respectively.
💡Of course, the power of a combination attack will be greater. For example, write a series of tweets on the same topic from multiple angles. Then use one main tweet as the "entry point," while adding links to other related content in the comments or at the top of your homepage, creating a chain of interconnected content. Series content and interlinked information increase the relevance of candidate tweets, not only expanding candidate generation and social graphs but also potentially triggering additional interaction behaviors (such as clicking into the homepage) to enhance interaction signals. Moreover, this content linkage can also trigger high interactions in the short term while forming a long-term content matrix, helping you maintain stable performance in rolling aggregation (50 days + short-term) statistics.
This series linkage strategy enriches the content ecosystem and aligns with the platform algorithm's measurement standards for interaction and relevance, thereby helping to enhance overall exposure.
Efficient Traffic Generation and Leverage Strategies
1⃣Heat Leverage: Quickly Responding to and Following Hot Events
During the data collection phase, the system incorporates the high-frequency interactions and user attention brought by hot events into "Tweet Engagement" and "User Data," treating the "freshness" or "timeliness" reflected in them as key features. Since these features help the algorithm assess the current heat and relevance of the content, tweets that timely follow hot topics are more likely to be prioritized in the candidate generation phase and continuously score higher in the "Heavy Ranker" due to the rising interaction volume, ultimately gaining higher exposure opportunities in the "Mixing" phase.
An effective method is to seize the immediacy of hot events by publishing relevant comments or insights at the first opportunity to take the lead. After posting, continue to update your views or supplement information based on the event's progress to ensure the tweet remains active. The system will combine interaction and comment data to assess the content's freshness, thus granting higher weight in subsequent scoring and recommendations, allowing your content to maintain an advantage throughout the recommendation process.
2⃣Controversy Leverage: Creating Controversy and Discussion Points
By presenting unique insights or controversial topics, you can quickly spark a large amount of discussion and replies, forming strong interaction signals. Especially when the topic attracts more KOLs to participate in retweets and comments, the tweet has the opportunity to gain broader user exposure. However, when using the controversy leverage, it is essential to ensure that discussions remain within a reasonable and rational scope to avoid negative feedback (such as hiding, blocking, reporting, etc.) due to violations or sensitive content. Because such negative operations can weigh as much as -74.0 to -369.0, potentially affecting not only the exposure of that tweet but also the entire account negatively.
3⃣Celebrity Leverage: Interacting with Hot Figures or Institutions
Interacting with hot figures or institutions, especially from high-influence accounts, will lead the system to determine that the account has a higher "TweepCred" and dissemination potential, which is directly reflected in the user's "Social Graph." Such interactions not only enhance the social value of the content itself but can also be further amplified through deep learning models, resulting in higher exposure rates in the final "Mixing."
Therefore, you can increase exposure by mentioning or @ing relevant hot figures or institutions, while timely interactions may attract their responses, facilitating secondary dissemination. This strategy helps push your content into a broader social network, further enhancing credibility and dissemination effects.
Summary
Maintaining stable high-quality output, adept at seizing short-term trends, utilizing diverse interactive techniques, and continuously tracking data while adjusting strategies in a timely manner will allow you to respond perfectly regardless of how the algorithm weights change. If you have other valuable insights, feel free to share them in the comments.
Thank you for reading this far! If you find this content helpful, feel free to follow my account @0x_kuma; likes, comments, and retweets will be the greatest encouragement for me!
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
