Skip to content

Inputs and Outputs¤

Auto-CORPus can be used directly on one file, or a list of files. However, to use it on a directory containing multiple files and nested directories, there is an assumed structure. This is useful for analysing an article along with its associated tables and supplementary information.

Input¤

Auto-CORPus processes two categories of biomedical literature: full text with tables and supplementary information.

Full text input files can be in HTML, XML or PDF formats, with an option to process standalone HTML files which describe a single table.

Supplementary information files can be in PDF, word processor (doc, docx), spreadsheet (xlsx, xls, csv, tsv) and presentation (pptx, ppt, pptm, odp) formats.

Auto-CORPus relies on a prescriptive directory structure and naming convention to distinguish and group full text, table and supplementary information files.

The input directory passed to Auto-CORPus should contain article full text files and optional subdirectories for tables and supplementary information for each article.

The article full text file is named:

  • {article_name}.html or {article_name}.xml or {article_name}.pdf

Article table HTML files are placed in a subdirectory named:

  • {article_name}_tbl

The table files in the subdirectory must contain _X at the end of the file name, where X is the table number.

  • {any_name_you_want}_X.html

Article supplementary information files are placed in a subdirectory named:

  • {article_name}_si

The supplementary information files in the subdirectory can have any name, but must contain a .pdf, .docx, .doc, .xlsx, .xls, .csv, .tsv, .pptx, .ppt, .pptm or .odp file extension.

Output¤

Auto-COPRus will write output to the “root” location passed to it, replicating the input directory structure.

Input Output 
inputs
├── PMC1.html
├── PMC1_tbl
│   ├── results_tbl_1.html
│   └── results_tbl_2.html
└── PMC1_si
    └── suppl_methods.docx

outputs
├── PMC1_bioc.json
├── PMC1_abbreviations.json
├── PMC1_tbl
│   └── PMC1_tables.json
└── PMC1_si
    └── suppl_methods.docx_bioc.json

inputs
├── article1.pdf
└── article1_si
    ├── extended_results.xlsx
    └── file3454.pdf

outputs
├── article1_bioc.json
├── article1_abbreviations.json
├── article1_tables.json
└── article1_si
    ├── extended_results.xlsx_tables.json
    └── file3454.pdf_bioc.json

For each full text file, a bioc and abbreviations json file is output to the root.

If a tables subdirectory was not given for the article, the tables json file is output to the root.

If a tables subdirectory was given, the tables json file is output to the subdirectory. The tables json contains all tables from the separate HTML files and any tables described within the main text.

The processed supplementary files are output to the supplementary subdirectory. If Auto-CORPus detects text in the input file, a bioc file is output with _bioc.json appended to the end of the original filename. If Auto-CORPus detects one or more tables in the input file, a tables json file is output with _tables.json appended to the end of the original filename. If both text and tables are detected, then both formats will be output.