Parse PDF with left-right format

[xyzhu68]

Hi,
I have a pdf file, whose content on each page is written in 2 columns. After Chunking, it seems that the parser try to read the article from left to right, ignore the space between the 2 columns (see the yellow highlighting). It shall be read first on the left side and then on the right side:

The chunk method used is "general", PDF Parser is "DeepDoc".

[noybiss]

use paper chunking method

[xyzhu68]

Hi,
I used paper to parse the pdf. But the result is worse than general. Some table or image cannot be related to text correctly. Maybe it is the pdf file, which is something special for your parser. If you are interested in it, here ist the link:
https://assets.stiebel-eltron.com/Celum/Docs/Dokument-PDF/DOC-00060361.pdf
Sorry it is in German and is an installation instruction of a heat pump.

[ZhenhangTung]

Hi, I used paper to parse the pdf. But the result is worse than general. Some table or image cannot be related to text correctly. Maybe it is the pdf file, which is something special for your parser. If you are interested in it, here ist the link: https://assets.stiebel-eltron.com/Celum/Docs/Dokument-PDF/DOC-00060361.pdf Sorry it is in German and is an installation instruction of a heat pump.

@xyzhu68 what about using manual chunking method?

[xyzhu68]

If I use manual method, the parser reads the document from left to right, ignoring the space in the middle.
If I use paper method, the left-right structure is repected. But the tables and figures are not related to the text correctly. I think, the reason is, there is no labels as "figure 1" or "table 1" in the text. The text immediate above or beneath is the related text. The parser has difficulty to handle it.