The Tender Discovery Service (TDS) aims at enabling customers to easily and quickly identify pertinent open public administration tenders, tailored to customer interests, rich company profiles and capabilities. Hence, TDS is aimed at marketing and sales lead in both large, and small and medium-sized companies (SME), facilitating business growth and search for prospective customers. It is being implemented as an offering of Cerved’s marketing solutions and lead generation portfolio of products.The business case also aims to serve companies across jurisdictions through a set of services provided by the BusinessGraph.
TDS supports companies across jurisdictions in discovering new untapped business opportunities in the Italian market. It enables easy, fast and intuitive discovery of relevant open public administration tender calls tailored to rich company and company-related information. TDS focus is on marketing solutions and sales managers.
Economic and Societal, Public Administration, SMEs, Companies
The customer segments in focus include both large corporate clients and SMEs/ small businesses that operate for or wish to extend their client base with public administrations. This amounts to potential client base of around a million legal entities in Italy, participating in open public tenders over the last 5 years and in the line of a hundred thousand unknown firmographics potentially coming from non-Italian jurisdictions. TDS in intended to extend the offering of Cerved’s marketing solution and lead generation portfolio of products.
TDS aims at enabling customers to easily and quickly identify pertinent open public administration tenders, tailored to customer interests and company profiles. The service is a truly data driven solution relying on interlinking rich company data with open and historical tender calls.
The main challenges this business case is set to solve are :
- Getting the right info for the customer i.e. corresponding to customers’ needs
- Assuring extensive tender coverage
- Harmonizing data coming from numerous sources
- Changing formats of tender calls from year to year requiring constant maintenance
- Inaccuracy in profiling based on companies declared economic activities due to staled information
- Offering service for non–Italian companies through cross jurisdictional company information
The TDS service is a set of algorithms and services that enables easy, fast and intuitive discovery of relevant open public administration tender calls based on tender call content, company profiles, and interests expressed through past participations to similar types of calls by customers. The current solution relies on open tender calls data and closed contract data for Italy, as well as the Italian company register. The final solution enables the user to authenticate, search, filter and rank relevant open tenders and receive recommendations based on their profiles. The recommendation service, which is the core part of the TDS service, relies on: (a) tender-to-tender similarity, (b) recommending open tender based on company’s past participation history, i.e. contracts, and (c) recommending open tenders to companies without tender participation history by extending the previous functionality through company-to- company similarity. This service is being delivered as part of Cerved’s Marketing Solutions lead generation offering.
The TDS team has focused on data enrichment and interlinking company and company related data:
- We developed a machine learning approach for better categorization of tender calls based on the title of call and going beyond common procurement vocabulary (CPV) tags.
- Named entities for places and keywords from tender content helped us redefine the content and improve geo referencing of tender calls.
- Finally we enrich the content of each tender call with a multi-label approach based on named entities and Italian data specific categories.
We advanced our tender recommendation approach by modeling the expected behavior of a company, i.e. moving from public tender to similar public tenders recommendation approach, to include proposing similar tenders using also past tender notices. The latter relying on historical tender participation and approximate k-nearest neighbors algorithm for building a forest of similarity trees.