Close
Type at least 1 character to search
Back to top
AIBOO, Artificial Intelligence

AIBOO, an artificial intelligence tool for business

Since its creation, innovation and development have been at the heart of the Maltemdigital strategy agency . And the group has taken a new step by creating its own Research & Innovation laboratory: Maltem Lab. Under the responsibility of Frédérick Bisone, its objective is to encourage research and innovation in the group's various areas of expertise.

The latest project, AIBOO, is an artificial intelligence tool based on Natural Language Processing (NLP). The objective of AIBOO is to automatically collect calls for tender, read them, understand them, and then determine and automatically extract the relevant content. Once this processing is complete, AIBOO is able to :

  • Automatically calculate a match score between a CV and a tender;
  • Determine market trends (relevant content that comes up regularly, that has disappeared, that is emerging...) by sector, by studying calls for tender over time;
  • Synthesise an invitation to tender with an automatic summary in order to improve the speed of understanding of it;
  • Assessing a CV against market trends.

How the tool works

Step1: Collecting tenders

AIBOO automatically collects the calls for tender issued by our customers via various web services. These extracted calls for tender are then stored with their various properties (content of the call for tender, date of collection, collection platform, etc.) and then converted into text format to enable artificial intelligence processing.

Step2: Pre-processing of tenders

The next step is tokenisation. Tokenisation seeks to transform a text into a series of individual tokens, each token representing a word. Word contractions can sometimes cause a problem during tokenisation. For example, the sentence "I ate an apple" would be tokenised as "I ate an apple", thus losing the meaning of the subject, in this case "I", which would be transformed into a j. Once the tokens have been obtained, we delete the most frequent words or stop words. Indeed, some words are found very frequently in the language (words such as "but, one, ...). Often, these words do not add any information to the meaning of the text. This is why we remove them from our word list.

 

🌎 Maltem Group is a digital strategy agency, enabling companies worldwide to develop their digital potential. Also an expert in cyber security, Maltem Group is your trusted partner. 🌎

Step3: Extraction of relevant data

There are many factors that prevent a simple top 10 list of the most relevant words from a tender. Indeed, for example, the fact that several languages may be used for tenders, or simply synonyms may bias this word by word analysis. For example, database developer and SQL programmer are two names that refer to the same concept (a person who develops scripts for databases). A word-by-word analysis would not understand that these two phrases have the same meaning. Therefore, once we have extracted the important tokens from the tenders, we replace these tokens with their concept in an ontology. An ontology is a structured set of terms and concepts linked by relationships. This allows us to get away from the problems of translation, synonyms and context.

Example: In a call for tender, the word bank is used. The use of ontologies allows us to replace the word bank with the concept bank of the ontology. In concrete terms, this allows us to replace the word with a unique concept identifier. Thus, the word bank in English would have the same identifier, thus avoiding translation problems. Moreover, the concept of bank is itself linked to other concepts in this ontology, in the case of the word bank, it is linked to concepts such as financial institution, company, bank account, banking system, deposit bank, financial sector...

The usefulness of concept relations is that finally it would be possible to link two concepts such as bank and credit, each of which is related to the financial sector, the bank account, etc.

Finally, in order to extract the relevant content of the calls for tender, we perform an extraction of the different concepts present in them.

Once these concepts have been extracted, it is possible to present the main areas of each tender and thus summarise its content.

Step4: Linking with CVs

The same concept extraction algorithms are used for CVs. Then, AIBOO's artificial intelligence algorithms automatically calculate a correspondence score between the concepts extracted from a CV and those of a tender in order to calculate whether or not a CV is relevant to a tender.

Step 5: Trend calculation

As tenders are extracted and processed at recurring periods, AIBOO stores the results over time and then provides an interface to view the evolution of trends by domain or technology over months or years.

Our objectives

  • Extract relevant information from a tender by producing a summary that is more easily understood by a sales team
  • To increase the capacity to process tenders
  • Save time for the sales teams by allowing them to free themselves from certain time-consuming tasks (complete reading of a call for tenders, verification of CVs compatible with the profile sought, etc.) and thus free up more time for responses
  • Facilitate the decision making of HR teams by matching a CV with market trends in a studied sector

In practice

  • A sales team handling large volumes of tenders wants to automate the first filtering of these. AIBOO allows the automatic collection of large volumes of tenders and thus proposes only relevant tenders, as well as a list of eligible CVs. The possibility of reading the automatic summary also means that no time is wasted on calls for tender that are not within the scope of the sales representative, and therefore saves him time.
  • An HR team would like to know about changes in its clients' sector in order to guide the company's future recruitment (presence or absence of emerging skills or, on the contrary, skills in greater demand).

Do you want to know more about AIBOO, the artificial intelligence? Would you like to be accompanied by the leading digital strategy agency? Contact Frédérick Bisone[/vc_column_text]

X