| """Stable public feature preprocessing for Forge2Vec.""" |
|
|
| from __future__ import annotations |
|
|
| from typing import Optional |
|
|
|
|
| GENRE_VOCABULARY = { |
| "Comedy": 1, "Fantasy": 2, "Romance": 3, "Action": 4, "Drama": 5, |
| "School": 6, "Shoujo": 7, "Shounen": 8, "Adventure": 9, |
| "Supernatural": 10, "Kids": 11, "Slice of Life": 12, "Sci-Fi": 13, |
| "Music": 14, "Historical": 15, "Mystery": 16, "Boys Love": 17, |
| "Adult Cast": 18, "Horror": 19, "Isekai": 20, "Sports": 21, |
| "Mecha": 22, "Psychological": 23, "Anthropomorphic": 24, |
| "Martial Arts": 25, "Girls Love": 26, "Super Power": 27, |
| "Mythology": 28, "Military": 29, "Avant Garde": 30, "Seinen": 31, |
| "Parody": 32, "Harem": 33, "Gourmet": 34, "Workplace": 35, |
| "Space": 36, "Ecchi": 37, "Award Winning": 38, "Detective": 39, |
| "Gag Humor": 40, "Strategy Game": 41, "Gore": 42, "Vampire": 43, |
| "Mahou Shoujo": 44, "Samurai": 45, "Idols (Female)": 46, |
| "Suspense": 47, "CGDCT": 48, "Educational": 49, "Villainess": 50, |
| "Love Polygon": 51, "Racing": 52, "Childcare": 53, "Iyashikei": 54, |
| "Team Sports": 55, "Combat Sports": 56, "Delinquents": 57, |
| "Idols (Male)": 58, "Medical": 59, "Memoir": 60, |
| "Reincarnation": 61, "Video Game": 62, "Magical Sex Shift": 63, |
| "Time Travel": 64, "Performing Arts": 65, "High Stakes Game": 66, |
| "Pets": 67, "Josei": 68, "Otaku Culture": 69, "Organized Crime": 70, |
| "Survival": 71, "Visual Arts": 72, "Erotica": 73, |
| "Reverse Harem": 74, "Romantic Subtext": 75, "Crossdressing": 76, |
| "Showbiz": 77, "Hentai": 78, |
| } |
|
|
|
|
| def genre_indices(names: list[str], max_length: int = 10) -> list[int]: |
| values = [GENRE_VOCABULARY.get(name, 0) for name in names[:max_length]] |
| return values + [0] * (max_length - len(values)) |
|
|
|
|
| def metadata_vector(content_type: str, year: Optional[int], score: Optional[float]) -> list[float]: |
| return [ |
| 1.0 if content_type.lower() == "anime" else 0.0, |
| 1.0 if content_type.lower() == "manga" else 0.0, |
| max(0.0, min(1.0, (year - 1960) / 70.0)) if year else 0.5, |
| max(0.0, min(1.0, score / 10.0)) if score is not None else 0.5, |
| ] |
|
|
