Sciling helps the Government of Spain to ensure open, effective competition in public procurement

When there are no bids in a public tender, this leads to a series of problems for the government administration.


  • Firstly, it entails a duplication of the human and economic resources and time spent on a failed tender, since it will have to be done again.
  • Secondly, such a void process means it is impossible for the public administration to carry out their activities planned or to provide citizens with the necessary services. This in turn creates great pressure to implement the subsequent processes within shorter deadlines, usually leading to less thoroughness and less suitable suppliers being selected.
  • Last but not least, a tender with no bids has a knock-on effect on starting up other projects that depend on it, causing delays.

What is more, tenders with no bids are only a part of the problem. According to a recent study on the inefficiencies in public procurement in Spain, there is only one single bidder for 40% of tenders. The lack of competition in such cases not only implies a less-than-perfect match between the project and the supplier; it also leads to a price rise of 20%. In a country in which public procurement accounts for 20% of GDP, this problem can be estimated as being worth around 10 billion euros a year.

The problem is possibly even more extreme in sectors with great demand in recent times. Such is the case in the cybersecurity sector, whose importance has grown increasingly as a result of the COVID-19 pandemic. In this context, INCIBE is very much aware of the need to implement all of its initiatives on time, in the right way and using the resources available.

About the client

The Spanish National Cybersecurity Institute (INCIBE) works together with the Secretary of State for Digital Transformation (SEAD) to bolster digital trust, improve cybersecurity and resilience and to contribute to the digital market in order to give the sector a boost and attract direct foreign investment.

The challenge

INCIBE is also troubled by this problem in the Spanish public administration, launching public tenders that relatively often end up with no bidders. Indeed, due to two tenders that received no bids, the agreement upon which this project is based had to be extended over time to be able to properly complete all of the activities planned.

Given this circumstance, INCIBE clearly saw the need to develop automated ways of matching supply to demand in the cybersecurity market. However, that is no simple goal. In order to match supply to demand in public procurement, it is first necessary to characterise the companies in the sector, taking into account the patents, products, services and information to be found in unstructured text data sources.

Given that all of the information necessary to carry out this task was to be found in unstructured text data sources, only the most powerful Natural Language Processing (NLP) would be able to take on the challenge. That is why it was decided to entrust this project to Sciling, which has been working for decades in groundbreaking academic research in the field, creating high-level NLP solutions for companies.

Strategic vision

In order to ensure the quality and suitability of the products to be developed, regular meetings were held between the two parties, with some of the end users present as well as the project managers. Hence it was possible not only to detect and correct any divergence between INCIBE’s expectations and the project’s true progress, but also to ensure the tool’s usefulness.

Our solution

Sciling has built a user-friendly visual dashboard that uses various modules to automatically classify and characterise tenders, companies and solutions, find the most noteworthy characteristics of the documents related to the cybersecurity companies and cross-reference documents from different corpora, visualising the relationship between them in graphs. Furthermore, in order to give it greater flexibility, Sciling has given the tool the possibility to specify filters to screen the data, enabling relevant documents to be selected.

While carrying out the project, Sciling assessed the performance of the different models for representing and classifying/labelling documents in order to find the most suitable combination between representation and classification. The result is a fast, precise and robust system that enables candidates to be prospected for a specific tender, giving INCIBE the means to avoid receiving no bids. Open, effective competition is thus ensured, together with more egalitarian, fair participation for SMEs, which is vital to maintain employment and sustainable development in Spain.

Moreover, the system also enables INCIBE to have at its disposal information about current trends in the cybersecurity sector, to better understand its structure and the companies involved, and even to detect new ones. All of this knowledge about the market in which the public administration carries out procurement enables it to do so more intelligently and efficiently.


New companies detected

The implementation process

In order to meet the ultimate goal of matching supply to demand in the cybersecurity sector, the project was divided into two sub-projects.

The first of these concentrated on characterising the sectors in the company via the unstructured text in their websites. To comply with this task, various modules were developed that carry out the functions of classifying, modelling topics, visualising and extracting information. These modules are useful tools to automatically identify the different characteristics of companies involved in cybersecurity, as well as the most relevant trends in the sector.

The second sub-project, based on what was achieved in the first one, concentrated on developing a tool to match tenders to companies. When there is a tender related to the cybersecurity sphere, the tool enables the most suitable companies to be found to carry it out so that bids are received. It can also work the other way around, finding suitable tenders for a company.

Technologies used

Natural Language Processing (NLP)

To process data

Word embeddings and Lateral Dirichlet Allocation (LDA)

To model topics

Algorithms based on semantic distances

To identify the relationships and distances between sources

Banana dashboard on Solr database

To manage, analyse and visualise the information

Why us?

  • We have wide-ranging experience in the most powerful Natural Language Processing techniques.
  • Our team is mostly made up of people with PhDs in IT.
  • We do continuous research to find how the most cutting edge techniques can be used to overcome business challenges.
  • We have over 15 years’ experience in carrying out projects involving research into automatic learning and its applications.