What is CV Parsing and how does it work?

Reading CVs is always on the agenda for those who occupy themselves with search and selection. But what happens when a computer reads a CV? While the recruiter can quickly and easily find the information that they consider relevant and required for a job posting, it still takes a lot of time to analyse every single CV. This is where CV parsing systems can play a fundamental role in the recruiting or talent acquisition process. Let’s find out more.

What is meant by CV / Resume Parsing?

CV Parsing, also known as Resume Parsing, is an extremely beneficial tool for those involved in search and selection. It is a process of extracting and converting unstructured data from a document – in this case a CV – into structured information, which is then automatically organised and stored within the relevant database. Parsing, automatic extraction and filing of data can be very useful in recruitment activities. A typical feature of the parser is to extract the content of a CV in “plain text” so that keywords can be searched within the document. The parsing system can also automatically fill out an application form meaning that recruiters can fill out many forms quickly. 

Traditional Parsing VS Intelligent Parsing

Compared to traditional parsing (simple digitisation, or text extraction), artificial intelligence has made the process more innovative by focusing on the extraction of structured information. An “intelligent” parsing system allows:

• the extraction of structured information, which can be directly entered into a database and used for much more precise, advanced and efficient searches than can be done with simple text;
• the layout analysis (Document Layout Analysis) to be improved specifically for CVs which, compared to other documents, have complex and diverse layouts.

Think of the analysis of a CV file, generally arranged in two or more columns. A generic or old-generation parsing software performs the extraction of simple, ordered text following the line of the document: since it is a text arranged in two columns, the result would be a sentence that is not consistent with the original meaning. The meaning of the text can be reconstructed with geometric analysis of the document, but it is of little use in keyword searches (as opposed to what happens with AI parsers). The new parsing system has the advantage of offering a complete understanding of the text. Furthermore, it is able to read the document, even from a scan or photograph, recognise and read sections and tables, as well as understand the meaning and relationships between the various parts of the text.

FEATURES	INDA	TRADITIONAL CV PARSING
1. Digitising of the CV
2. Standard file forms accepted (pdf, doc, docx, txt)
3. Any text file accepted (incl. odt, rtf, html, pptx)
4. Image files accepted (jpeg, jpg, tif, tiff)
5. Correct processing of CVs, even with complex layout
6. Automatically recognising pertinent information (skills, job title, level of education etc)
7. Organising information into a structured format suitable to insert into a database

N.B. In order to accept any type of file in textual format, the Inda parser carries out a preliminary conversion step, returning the text to standard format [3]. To process image type files, Inda provides an Optical Character Recognition (OCR) step [4]; for CVs with complex layout (many sections, multiple columns, graphs, etc.), there is Document Layout Analysis specialized for CVs with even the most complex formats [5]. The Inda parser also includes a Named Entity Recognition (NER) step which, through specially trained and focused neural networks, is able to recognise all the salient information in a CV for a recruiter [6]. Finally, a Relation Extraction (RE) step links the information extracted from the CV and allows it to automatically obtain structured data [7].

How does CV parsing work?

The new parsing systems use specific artificial intelligence (AI) techniques to recognise and extract text and information from a document (CV). To better understand how it works, let’s consider the specific case of Inda (INtelligent Data Analysis), the AI ​​solution for the HR world, equipped with a parsing system designed specifically for CV analysis. This system uses computer vision and Natural Language Processing (NLP) techniques to recognise the file structure and the graphic layout, and to analyse human language, extract text and, finally, map and store useful data. All with the aim of finding the most qualified and suitable candidates for the job posting.

To start the parsing process, you begin by uploading a file, the CV. As we know, there are many different types of CV layout and format. Generally, it is the recruiter who reads, analyses and contextualises the information. In the case of parsing, with self-learning  AI (machine learning), the system is able to read and recognise the different formats (over 12 CV file extensions for Inda) by analysing their layout, sections, structures.

Following the recognition of text and images in the file, the actual analysis and extraction of information from the CV takes place. The personal information, professional experience,  education and skills are all extracted. This information will be used for self-completion of the application form or the automatic compilation of the candidate’s application. Currently, many companies still ask candidates to fill out lengthy application forms with the risk of losing  talent along the way. The intelligent parser tries to avoid this problem by automatically completing application forms.

Reading a CV: a test for artificial intelligence

It is possible that the use of parsers in the recruitment process can raise questions or doubts about the accuracy of the document structure and language recognition. It is true that there are numerous CV formats and that the style of language can vary greatly. For example, how would artificial intelligence distinguish between the name and surname of the candidate, and, for example, the company name? Also, would a date be interpreted correctly depending on how it is formatted? Ultimately, how is all of the information interpreted? These are legitimate questions and artificial intelligence systems are combatting those challenges with techniques (DLA, OCR, NER, etc.) capable of achieving valid and accurate results.

Why is it useful to use CV parsing in recruitment?

Artificial intelligence applied to the human resources sector can provide the recruiter with the tools and support necessary to perfect the search and selection process. More and more often artificial intelligence is chosen to integrate with recruitment software or an Applicant Tracking System (ATS).

In this way, some of the candidate’s activities, as well as the recruiter’s activities are simplified. With CV Parsing, artificial intelligence reads the CV, identifies key information and automatically completes the application form, making the application experience more positive. In addition, the ability of AI to distinguish images from text, recognise and extract skills and also fill in the drop-down menus, speeds up the addition of a new application in the system, improving the quality of the recruiter’s work.

The advantages of CV Parsing

The advantages of artificial intelligence are numerous and apply to various stages of the recruitment process. Specifically, CV parsing allows you to:

• Analyse different types of CV files, including scans or images of the CV
• Speed up the activity of reading and analysing information
• Collect and organise large amounts of data in digital form
• Improve the search for keywords contained within the CV
• Spend less time doing manual tasks
• Speed up the creation and insertion of new applications in the database
• Speed up the selection process
• Improve the quality of the recruiter’s work
• Make the candidate’s application experience more positive 

Other uses of CV Parsing

CV archives are often composed of digital files, saved in folders on company PCs or servers, plus CVs that are kept in paper format. Saved as such, these documents are not very useful as they do not have any structured information (e.g. name, surname, email, etc.) that can easily be accessed. If all of these CVs were to be processed by a special parser, there would instead by a list of structured information that would be of enormous value. This data, in fact, could be imported into an Applicant Tracking System (ATS) or help create a database of candidates.

Visit the website for more details on Inda parsing or to learn more about how Inda’s document parsing can be used in other ways.