{"blocks": [{"key": "e7e3ec7f", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "81c7ef1d", "text": "Choosing technical approaches for DOT, Meta’s bot-detection tool aimed at finding malicious duplicated content.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "caf688d9", "text": "Question", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "375e653a", "text": "What models or algorithms could help identify malicious duplicated content, and why are they suitable?", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "36ae8444", "text": "Hints", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "6828a705", "text": "Discuss text hashing, TF-IDF cosine similarity, embeddings with Siamese networks, transformer encoders, clustering thresholds, and supervised vs unsupervised approaches.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}], "entityMap": {}}