Intel’s newest open-source project and addition to their AI offerings is… Polite Guard.
Intel Polite Guard is an open-source natural language processing (NLP) language model for classifying text if it’s polite or not.
Intel Polite Guard builds off BERT for text classification and this NLP model will classify text as either being polite, somewhat polite, neural, or impolite. Intel has made their data-sets and source code to Polite Guard open-source on GitHub and Hugging Face. As part of this is also a politeness benchmark for being able to compare future NLP models based on their politeness classification.
Intel is hoping Polite Guard can help “significantly boost customer satisfaction and loyalty” for services employing the NLP model to enhance the customer experience.
Intel Polite Guard is under the MIT license. Those wishing to learn more about Polite Guard can do so via the Intel Community blog. Those wanting to jump right in to this open-source Intel project can find the Polite Guard assets on GitHub.
