Tell whether two product records are the same item

Fields

• Incoming offer, each column as column: value
• Existing product page, each column as column: value

Questions

1. Do both records say enough about the product to judge whether they match? Yes / No

   Read both records. Yes: Each record names the product clearly enough, through its title, brand, model or description, to compare it with the other. No: One record is empty, a placeholder, or too vague to tell what product it is, such as a title that is only a category name.

2. How closely do the two records match under the matching rules? Scale

   Read both records and the matching rules in the context. Judge brand, model, variant, condition and contents. Leave numbers out: code compares size, capacity, weight, wattage, pack count and price.

   • Different products A different brand, or a different model or product line.
   • Related offer The same brand and model or line, but a different variant, condition or contents.
   • Same product The same brand, model or line, variant, condition and contents, however they are worded.

3. Do the two records name the same brand? Choice

   Compare the brand or maker each record names, in any column.

   • Same brand Both records name the same brand or maker, however it is written, such as with "and" for "&" or in capitals.
   • Different brands The records name different brands or makers.
   • Not stated At least one record names no brand or maker.

4. Do the two records name the same model or product line? Choice

   Compare the model codes and product line names in both records, in any column.

   • Same model The same model code or product line, ignoring spacing, capitals, hyphens and a brand prefix on the code.
   • Different model Different model codes or lines, including a different generation or version of the same line.
   • Not stated At least one record gives no model code or product line.

5. Do the two records describe the same variant of the product? Choice

   Compare the colour, flavour, finish, edition or style each record gives. Names that mean the same shade or style, such as "navy" and "navy blue", are the same.

   • Same variant Both records give the same colour, flavour, finish, edition or style.
   • Different variant The records give a different colour, flavour, finish, edition or style.
   • Not stated At least one record gives no variant.

6. Does one record describe a used, refurbished or open-box item and the other a new one? Yes / No

   Read the condition, title and notes of both records. Yes: One record describes the item as used, refurbished, renewed, open-box or ex-display, and the other as new. No: Both records describe the same condition, or neither says.

7. Does one record include an item or accessory that the other says is left out? Yes / No

   Read what each record says is in the box, included or not included. Yes: One record lists an accessory, part or extra product that the other says is missing, not included or sold separately, or one is sold as a bundle or set with an item the other plainly lacks. No: The records describe the same contents, or neither says what is included, or one simply lists less detail without saying anything is left out. Different words for the same part, such as "paddle" and "flat beater", are the same part.

Lens columns

enough_information, enough_information_probability, match_level, match_level_average, brand_match, brand_match_probability, model_match, model_match_probability, variant_match, variant_match_probability, condition_differs, condition_differs_probability, bundle_differs, bundle_differs_probability