> For the complete documentation index, see [llms.txt](https://docs.ggwp.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.ggwp.com/api-docs-chat.md). # API Docs: Chat ## **Introduction** GGWP provides an API you can use for processing chat data and integrating actions in your chat application workflow. This product is designed to identify relevant events at the message level, as well as track the evolution of a conversation as a whole and the contributions of each individual user. GGWP can detect incidents (positive and negative), analyze a conversation's temperature and generate user reputation scores, among other things. ## Onboarding Guide Before starting to use the API, it is important to understand how this product works in order to unlock its full potential. ### **Understanding GGWP’s Chat Moderation** GGWP's Chat Moderation service covers 3 use cases: * Real-time content filtering - the API allows you to create custom settings for severity and topic selection so that you can block the types of content you want to prevent in your specific community. * Conversation analysis - designed to give you a cumulative view of positive and negative incidents in a session using contextual elements of the conversation, combined with GGWP’s user reputation system. This can be integrated with different types of warnings or alerts on the client side. * Automod service - allows you to automate the sanctioning of toxic players through mutes, suspensions, and bans so that you can give offending players enough time to cool off and prevent further impact to other players. This system takes into consideration incident severities, detection confidences, and player history and session behavior in order to determine when and for how long to punish a player. You will make 1 API call for each individual message or block of messages, including some basic information to identify the sender and session (more info in the “Chat API” section below). GGWP takes care of all the processing steps in order to provide the relevant insights. Once the response is sent back to you, we recommend integrating different actions to improve the user experience on your platform. Some example actions could be replacing the original message with our filtered version, sending user warnings or muting the most disruptive participants. ### **Configurable Settings** Custom settings can be defined in order to select the types of content you want to prevent in your platform. This can be done in 2 ways: * At onboarding time (preferable) - you share the necessary configurations with your Account Manager and GGWP incorporates them in your specific API settings. * During each API call - through the use of headers, you can also input some custom settings every time you make an API call (see “Chat API” section for details). This has some limitations since not all custom elements can be provided and is more prone to error. In addition, if you would like to enable GGWP’s Automod service, please contact your GGWP Account Manager. Here is a list of things that can be configured: * **Topics**: Available topics include Violence, Sexual Content, Verbal Abuse, Identity Hate, Profanity, Link Sharing, Drugs, Spam and Self-Harm. * **Filter Strength**: Filter strength configurations in the Chat API are based on a parent lens of safety. There are typically 4 values that can be selected for each topic, with a few exceptions. These 4 values are: * Filter strength = “Off” - the filter is not active, so all forms of toxicity are allowed. Suitable for environments with adults where we don’t want to impose any restrictions. * Filter strength = “Low” - only filters extremely offensive content, but allows for moderate and low forms of toxicity to go through. Suitable for environments with adults only. * Filter strength = “Medium” - filters out extreme and moderate forms of toxicity, only letting through mild toxicity. Suitable for environments with teenagers where some innocuous toxicity can be tolerated. * Filter strength = “High” - filters out all forms of toxicity. Suitable for environments with presence of kids. * **Custom Content:** If you have a custom allowlist, blocklist or replacement phrase list (phrases to use in substitution of toxic messages) that you would like to incorporate into your settings, please contact your GGWP Account Manager to share the files with us. ### **Language Support** GGWP's Chat Moderation service offers support for the following languages: * Arabic * Chinese (simplified and traditional) * Dutch * English * Filipino (Tagalog) * Finnish * French * German * Greek * Hebrew * Hindi * Indonesian * Italian * Japanese * Korean * Persian * Polish * Portuguese * Russian * Spanish * Thai * Turkish * Vietnamese New languages are added regularly. Please reach out to your Account Manager to learn about upcoming languages or submit additional languages for consideration. ### **Model Support** The categories covered by GGWP's Chat Moderation are described in the [model glossary section](/text-and-voice-model-glossary.md).