This Java project can be used to process in an efficient concurrent manner a set of PDF in a given directory by the GROBID service. Results are written in a given output directory and include the resulting XML TEI representation of the PDF.
You need first to install and start the grobid service, latest stable version, see the documentation. It is assumed that the server will run on the address http://localhost:8070. You can change the server address by editing the file grobid-client.properties.
Build the client:
mvn clean install
Parameters:
-
-in: path to the directory containing the PDF files to process -
-out: path to the directory where to put the results -
-exe: GROBID service to be used, one of[processFulltextDocument, processHeaderDocument, processReferences](defaultprocessFulltextDocument) -
-n: concurrency for service usage (default 10)
For example:
java -jar target/org.grobid.client-0.5.4-SNAPSHOT.one-jar.jar -in ~/tmp/in2 -out ~/tmp/out -exe processFulltextDocument
This command will process all the PDF files present in the input directory (files with extension .pdf only) with the processFulltextDocument service of GROBID, and write the resulting XML TEI files under the output directory, reusing the file name with a different file extension (.tei.xml), using the default 10 concurrent workers.
java -jar target/org.grobid.client-0.5.4-SNAPSHOT.one-jar.jar -in ~/tmp/in2 -out ~/tmp/out -n 20 -exe processHeaderDocument
This command will process all the PDF files present in the input directory (files with extension .pdf only) with the processHeaderDocument service of GROBID, and write the resulting XML TEI files under the output directory, reusing the file name with a different file extension (.tei.xml), using 20 concurrent workers.
Full text processing of 136 PDF (total 3443 pages, in average 25 pages per PDF) on Intel Core i7-4790K CPU 4.00GHz, 4 cores (8 threads), 16GB memory, n being the concurrency parameter:
| n | runtime (s) | s/PDF | PDF/s |
|---|---|---|---|
| 1 | 210.9 | 1.55 | 0.64 |
| 2 | 117.2 | 0.86 | 1.16 |
| 3 | 85.5 | 0.63 | 1.59 |
| 5 | 69.8 | 0.51 | 1.95 |
| 8 | 60.9 | 0.45 | 2.23 |
| 10 | 60.2 | 0.44 | 2.26 |
As complementary info, GROBID processing of header of the 136 PDF and with n=10 takes 3.82 s (more than 15 times faster than the complete full text processing because only the two first pages of the PDF are considered), 35.60 PDF/s. In similar conditions, extraction and structuring of bibliographical references takes 27.9 s, 4.87 PDF/s.
Benchmarking with more files (e.g. million ISTEX PDF). Also implement existing GROBID services for text input (date, name, affiliation/address, raw bibliographical references, etc.). Better support for parameters (including elements where to put coordinates).
Distributed under Apache 2.0 license.
Main author and contact: Patrice Lopez (patrice.lopez@science-miner.com)