Ask HN: Algorithm for searching text with similar meaning
For example, I have a long text that says, "I have to change the screw number 1234," and a short one that says, "screw number 1234 changed." Both were inserted by different people, referring to the same thing. I thought of using an LLM (GPT-4), however, my dataset is too large (millions of entries) and it would be expensive. Is there any other better or good enough way? Thank you.
阅读原始来源
来源记录
指标快照
points · 1 votes
2026-11-05 16:03:07.955Z
cumulative since publication
研究分析
用户报告的问题
来源描述：For example, I have a long text that says, "I have to change the screw number 1234," and a short one that says, "screw number 1234 changed." Both were inserted by different people, referring to the same thing. I thought of using an LLM (GPT-4), however, my dataset is too large (millions of entries) and it would be expensive. Is there any other better or good enough way? Thank you.
频次与替代方案
记录作者：mococa。 来源票数: 1 votes (2026年11月5日 UTC; cumulative since publication)；前次同口径记录为 1（2026年11月5日 UTC），两个观测值不等于需求增长。 可供比较的近期同类记录：“DeepSeek-V4: beam search leaves HCA/CSA compression caches on the wrong beams” (GitHub Issues, 2026年11月4日)；“[问与答] 纯工具性的软件创业走到头了” (V2EX, 2026年11月2日)。主题相近仅用于导航，未核验其是否属于同一需求或独立证据。
需求判断
将报告的任务或障碍作为访谈线索。功能请求、抱怨或替代方案搜索本身并不能确定预算负责人或付费意愿。
商业证据缺口
待补证据：受影响的独立用户、重复发生场景、当前绕行方案、实际时间或金钱成本、采购权限，以及对具体报价的响应。
需求验证动作
按相同任务整理原始问题，去重作者和转载；核对今天如何处理、实际成本多少，再测试明确报价后判断付费需求。