Odnoklassniki Introduces a Machine Learning Model to Reduce Unwanted Contacts and Communications
The Odnoklassniki social network has rolled out a machine learning driven system designed to curb unauthorized dating and unsolicited messages on the platform. This new approach aims to reduce friction between users who prefer to keep their online experience more private and protected from persistent or intrusive outreach.
Users who frequently decline invitations to chat or date will notice fewer alerts about activities from accounts that push for online conversations. The system limits the number of conversations initiated by such accounts with users who are not in their established circles. This helps ensure that people can browse, connect, and engage more comfortably without being overwhelmed by unwanted outreach.
To safeguard members, Odnoklassniki analyzes real time social interactions between unfamiliar users. It considers a wide range of signals, including friend and friend request patterns, page visits, the sending of gifts, and other engagement indicators. A neural network processes these signals to categorize accounts and identify potential harassment patterns. The model commonly flags behaviors such as adding strangers as friends in bulk, sending gifts to unknown users in large quantities, offering tutoring to strangers, and sending unsolicited personal messages. In many cases, users respond by declining friend requests, blacklisting certain accounts, refusing gifts, or reporting unusual activity. This helps the platform learn which profiles and actions are more likely to be intrusive, enabling smarter moderation.
Based on accumulated data, Odnoklassniki adjusts interactions between these groups of accounts and with remote contacts by narrowing notification visibility. Notifications about friend requests, group classes, gifts, or page visits on photos and posts can appear less frequently to unknown interlocutors. The model also reduces the volume of direct conversations between such user groups, keeping the platform a more welcoming space for genuine connections. In practice, the system limits the number of conversations that initiate private messaging with large-scale outreach by unfamiliar accounts.
During extensive testing conducted across diverse user segments over several months, the platform observed a meaningful drop in intrusive attention. Complaints about unsolicited contact from unfamiliar accounts decreased by a notable margin in comparison with earlier periods, indicating a real improvement in user comfort and perceived safety. The testing results support the conclusion that automated moderation can effectively balance openness with personal boundaries on a large social network. These outcomes were reported by independent observers who analyzed the deployment and its impact on user experience. [OK Lab] These findings align with a broader industry trend toward empowering users with clearer controls and smarter, privacy-respecting features that preserve the social nature of online communities without compromising personal autonomy. [Research Note]
The deployment of this machine learning technology is intended to make interactions on Odnoklassniki more pleasant and respectful. By reducing unwarranted interruptions, the platform seeks to maintain a friendly atmosphere while enabling meaningful connections for those who wish to engage. The approach reflects an ongoing commitment to user safety, informed by continuous feedback and data-driven refinement. [User Safety Report] The system is designed to adapt over time, learning from new patterns of interaction and adjusting moderation thresholds as needed to support a balanced and welcoming environment. [Community Review]