TrueMeans.World 如何判断一段内容“真正想表达什么”。
How TrueMeans.World judges what a piece of content really means.
意图分析指从一段内容推断说话者真正想达成的目的——推销、说服、塑造文化、转移话题、施加压力等——而不只是读字面意思。
Intent analysis infers the speaker's real goal behind a statement — promoting, persuading, shaping culture, deflecting, applying pressure — beyond the literal words.
所有问题都针对“这段内容本身”。我们不评判提问者是谁，也不猜测你的动机。
Every question targets the content itself. We don't judge who is asking, nor guess your motives.
判定由 TypeSafe 的 Jev 模型给出。它不输出“是/否”的结论，而是概率分布：是非题给“为真概率”，多选题给各选项概率，评分题给各档位概率与加权分值。
Verdicts come from TypeSafe's Jev model. It returns probability distributions, not conclusions: a probability of YES for yes/no, per-option probabilities for choice, per-level probabilities plus a weighted score for rating.
是非（noul）、多选（choice）、评分（score）。评分题的加权分值：
Yes/No, Choice, and Score. For rating:
score = Σ(level index × probability)，满分 = 等级数 − 1。, max = number of levels − 1.
涉及具体人物 / 事件、可被公开信息核实的说法时，会先检索公开信息再给概率；个人观点、情绪、处境类内容则不联网。
For verifiable claims about people or events, we search public sources before giving probabilities; personal opinions, emotions, or situations don't trigger search.
Jev 的主语言是英文，中文（CJK）精度较低；它不会算数、不会比日期，确定性计算都放在代码里。所有输出都是概率，不是结论，请结合公开信息自行判断。
Jev's primary language is English (lower accuracy on CJK). It can't do arithmetic or compare dates; deterministic math lives in code. All outputs are probabilities, not conclusions — verify with public information.
回到首页试用 → Try it →