!! This is more an issue to experiment with heuristics. How can a machine with "traditional" (non-AI) code create useful information. !!

When importing the PDF se2paper.pdf

one gets following BibTeX entry

@InProceedings{How,
  author   = {On How and We Can and Teach and Exploring New and Ways in and Professional Software and Development for Students},
  title    = {Microsoft Word - ieee_on_how_we_teach_jul_01.docx},
  abstract = {— Requirements and approaches for introductory
courses in software development at universities differ II. SETTING THE STAGE: SOFTWARE DEVELOPMENT AT 
considerably. There seems to be little consensus on which HDM 
languages are a good fit, which methodologies lead to the best 
...
  file     = {:C\:/Users/koppor/Downloads/se2paper-1.pdf:PDF},
}

However, the title should be better:

The properties of the file show

Tasks:

1. If title is "better" from the text importer, the title from the properties should not be used (class org.jabref.logic.importer.fileformat.PdfMergeMetadataImporter#importDatabase(java.nio.file.Path))
2. Improve abstract parsing. Maybe stripper.setSortByPosition(true); needs to be removed from org.jabref.logic.importer.fileformat.PdfContentImporter#getFirstPageContents. Maybe, two methods need to be done to be able to parse the title (depending on positon) and parsing the abstract (more on content)

Hint:

• Rely and add test cases to org.jabref.logic.importer.fileformat.PdfContentImporterTest and ´org.jabref.logic.importer.fileformat.PdfMergeMetadataImporterTest`