Data from the paper "PONTE: An NLP Pipeline for Transparent and Efficient Procurement"

This repository contains the data and models supporting the analyses presented in the paper "PONTE: An NLP Pipeline for Transparent and Efficient Procurement". It includes: Basic metadata directly downloaded from PLACE (ID, title, link to tender, CPV), when present. Metadata indicating, where download was possible, whether the technical and administrative PDF documents (on which the augmented metadata is based) were successfully downloaded. Augmented metadata generated through PONTE: objective of the contract, additional criteria (award and solvency criteria and special conditions), and CPV extracted at the levels 5 and 8. Lemmatized text used for topic modeling training and contextualized embeddings. Topic models trained with high (16) and low (8) granularities for the dataset whose CPV at level 2 is 79 (Business Services).