Wanting more details on data preprocessing

[Sylvia-Huang1997]

Excuse me. Could you please post more details on data preprocessing in README? I was puzzled about How you process source data into .txt.gz formats in folder data. It would be great if you could share the specific meaning of each field and the detailed process of preprocessing the data. Thank you so much！

[giacoballoccu]

There are entities and relations files.
The entities file have as a name nouns like brand.txt.gz, users.txt.gz, vocab.txt.gz etc It's a list of entities for that category e.g. Prada is in brand.txt

The relationship files have as a name a verb like also_bought or also_viewed etc. or they may have the form *_entityh_entityt like in the case of category_p_c.txt.gz where the first letter after the underscore represent the head entity and the second is the tail in this case (product -> belong_to -> category).
Relations files are formatted in a way such every line represent the product id (as you can see relation files have all the same number of lines) and for every line, there is the id of the entity that has as a tail.

e.g. extracted from category_p_c
line 0 0 20 18 19 4
line 1 0 100 166 46
This means that product 0 belong to category 0 20 18 19 4 and product 1 belongs to category 0 100 166 and 46.

[zhichaoxu-shufe]

in the category_p_c file, all line starts with 0; I think the line number already indicates the product_id

[giacoballoccu]

Usually, category 0 is the main category e.g. for dataset cellphones the entity category 0 is "cell, phone and accessories" so we can safely say that every item belongs to that category. So as I said in the precedent comment:
line 0 0 20 18 19 4
line 1 0 100 166 46
This means that product 0 belong to category 0 20 18 19 4 and product 1 belongs to category 0 100 166 and 46.