A Rule-driven Approach for Defining the Behavior of Negotiating Software Agents
Bibliographic record
Abstract
One problem with existing agent-mediated negotiation systems is that they rely on ad hoc, static, non-adaptive, and hardcoded schemes to represent the behaviour of agents. This limitation is probably due to the complexity of the negotiation task itself. Indeed, while negotiating, software (human) agents face tough decisions. These decisions are based not only on the information made available by the negotiation server, but on the behaviour of the other participants in the negotiation process as well. The information and the behaviour in question are constantly changing and highly uncertain. In the first part of the paper, we propose a rule-driven approach to represent, manage and explore negotiation strategies and coordination information. For that, we divide the behaviour of negotiating agents into protocols, strategies and coordination. Among the many advantages of the proposed solution, we can cite the high level of abstraction, the closeness to human understanding, the versatility, and the possibility to modify the agents' behaviour during the negotiation process. To validate our solution, we ran many agent tournaments, and used the rule-driven approach to implement bidding strategies that are common in the English and Dutch auctions. We also implemented simple coordination schemes across several auctions. The ongoing validation work is detailed and discussed in the second part of the paper. Un des inconvénients qu'on retrouve fréquemment dans les systèmes de négociation par agents est qu'ils reposent sur des schémas ad-hoc, non adaptatifs et figés dans le code pour représenter le comportement des agents. Cette limitation est probablement due à la complexité de l'activité de négociation elle-même. En effet, au cours de la négociation, les agents logiciels (humains) ont des décisions difficiles à prendre. Ces décisions ne sont pas seulement basées sur l'information disponible sur le serveur de négociation, mais aussi sur le comportement des autres participants durant le processus de négociation. L'information et le comportement en question changent constamment et sont très incertains. Dans la première partie de l'article, nous proposons une approche à base de règles pour représenter, gérer et explorer les stratégies de négociation ainsi que l'information de coordination. Parmi les nombreux avantages de la solution proposée, on peut citer le haut niveau d'abstraction, la proximité avec la compréhension humaine, la souplesse d'utilisation et la possibilité de modifier le comportement des agents durant le processus de négociation. Pour valider notre solution, nous avons effectué plusieurs tournois entre agents et utilisé l'approche à base de règles pour implémenter des stratégies simples applicables à l'enchère anglaise et à l'enchère hollandaise. Nous avons aussi implémenté des schémas simples de coordination impliquant plusieurs enchères. Le travail de validation, en cours, est détaillé et discuté dans la seconde partie de l'article.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.002 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".