ColumnBridge: local embeddings that stop before renaming the wrong CSV column

说明与用途

ColumnBridge 使用本地嵌入来建议 CSV 列映射，需人工审核后才重命名，防止数据放置错误。

问题

CSV imports can syntactically match but place data in incorrect columns, e.g., billing address vs shipping address both close to address, leading to business errors.

结果

ColumnBridge successfully identified correct column mappings for synthetic CSVs, flagged collisions, and required explicit approval before renaming. It ran locally in 8.58 s for first download and 0.35 s for cached offline runs, using ~198 MiB RSS.

工作流

• 1. Run ColumnBridge to generate a mapping report and review template with approved:false.
• 2. Inspect suggested mappings, review collisions and margins.
• 3. Approve or modify the plan in the review template.
• 4. Execute the apply command to generate a new CSV with approved mappings.
• 5. Verify that row values remain unchanged and no duplicate output names were created.

证据

ColumnBridge is a small local tool for people who repeatedly import contact lists or customer exports. It uses an open-weight sentence-embedding model to propose column mappings, then stops for a review before it renames anything.

来源

The first operation creates a report plus a review template with approved: false. Applying a reviewed plan is a separate command.

来源

One recorded first run, including the initial model download, took 8.58 seconds and used about 198 MiB maximum RSS on this Mac.

来源