This deposit contains benchmark code, data and results to assess the Python software time-agnostic-library v4.3.0. Two benchmarks were designed, one on the execution times and the other on the RAM. The goal is to assess whether time-agnostic-library is efficient and operable despite working live and without pre‑indexing. Moreover, all benchmarks are performed on four different triplestores: Blazegraph, GraphDB Free Edition, Apache Jena Fuseki, and OpenLink Virtuoso. The dataset used for the benchmarks contains bibliographical information about scholarly works in the journal Scientometrics only if the DOI is known. The data was extracted via Crossref. It is a temporal dataset in which provenance information and change-tracking have been managed by adopting the OpenCitations Data Model. Moreover, the dataset contains information on all the cited academic works. Journals, bibliographic resources, and authors always appear unambiguously, without duplicates. Finally, heuristics have been applied to recover the DOI of the cited works in case Crossref did not provide such information. In order to reproduce the results, extract the reproduce_results.zip archive. Then, execute run_benchmarks.sh on Linux or Mac, while run_benchmarks.bat on Windows. The results contained in results.zip were obtained using the following hardware specifications: CPU: Intel Core i9 12900k RAM: 128 GB DDR4 3200 MHz CL14 Storage: 1 TB SSD Nvme PCIe 4.0