On Validation and Predictability of Digital Badges’ Influence on Individual Users

Authors: Tomasz Kuśmierczyk, Kjetil Nørvåg

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Outcomes obtained from an evaluation on synthetic datasets and experiments on two badges from a popular Q&A platform confirm that it is possible to validate, characterize and to some extent predict users affected by the badge. empirical evaluation using synthetic data case studies of two badges from a popular Q&A platform
Researcher Affiliation Academia Tomasz Ku smierczyk, Kjetil Nørv ag Norwegian University of Science and Technology {tomaszku,noervaag}@ntnu.no
Pseudocode No No structured pseudocode or algorithm blocks (e.g., explicitly labeled 'Pseudocode' or 'Algorithm X') were found. The paper describes procedural steps in paragraph text and numbered lists.
Open Source Code Yes The code used in this paper is publicly available 4. 4http://github.com/tkusmierczyk/badges2
Open Datasets Yes In the real-data experiments we used a Stack Overflow dataset 5, that contains timestamped events from between July 2008 and September 2014 and some basic information (profile and actions record) about registered users. 5https://archive.org/details/stackexchange
Dataset Splits No The paper discusses evaluating for 'user validation problem' and 'user prediction problem' on synthetic data, and filtering users in real data, but does not provide specific train/validation/test dataset splits (e.g., percentages, sample counts, or explicit cross-validation schemes) for model training.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory amounts, or cloud instance types) used for running the experiments are mentioned in the paper.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers like Python 3.8 or CPLEX 12.4).
Experiment Setup No The paper describes parameters for synthetic data generation (e.g., Δλ, Δx, A, π) and statistical parameters for the semi-supervised clustering (e.g., FPR=0.25, FNR=0.4, σ=1.0). However, it does not provide specific experimental setup details such as hyperparameter values (e.g., learning rate, batch size, epochs) or system-level training configurations for model optimization.