Toolkits
This page is maintained by Nolwenn Bernard and Krisztian Balog. We welcome suggestions via email.
Search Engines
Toolkit | Paper | Summary |
---|---|---|
cwl_eval | Azzopardi et al., SIGIR 2019 | Implements various metrics within the C/W/L framework regarding the predicted user interactions with the ranked list of search results. |
SimIIR | Maxwell and Azzopardi, SIGIR 2016 | Framework for simulating users based on the Complex Searcher Model, including the actions of querying, snippet examination, relevance assessment, and stopping. |
SimIIR 2.0 | Zerhoudi et al., CIKM 2022 | Extends the original SimIIR framework by adding a dynamic query generator, introducing user types with distinct search behaviors, and using Markov models for stopping and query generation decisions based on these user types. |
SimIIR 3.0 | Azzopardi et al., SIGIR-AP 2024 | Extends SimIIR 2.0 with a conversational search workflow, Markovian users, cognitive states, and LLM-based components. |
Recommendation
To be populated
Conversational Information Access
Conversational Information Access might be seen as a specific type of Task-Oriented Dialogue (TOD), where user goals include search, recommendation and exploratory information gathering.
Toolkit | Paper | Summary |
---|---|---|
ConvLab-3 | Zhu et al., EMNLP 2023 | A dialog system platform that implements transformer-based (TUS and GenTUS) and LLM-based user simulators for task-oriented dialogue. |
CoSearcher | Salle et al., IRJ 2022 | Stochastic user simulator for conversational search refinement, modeling cooperativeness and patience of users. |
GenIRSim | Kiesel et al., CLEF 2024 | Implements an LLM-based simulator for the evaluation of generative information retrieval systems. |
iEvaLM | Wang et al., EMNLP 2023 | Provides the implementations of a configurable LLM-based user simulator and multiple conversational recommender systems for evaluation purposes. |
PyDial | Ultes et al., ACL 2017 | Multi-domain statistical spoken dialogue system toolkit with a user simulation component that operates on the semantic level for training and evaluating reinforcement learning-based algorithms. |
UserSimCRS | Afzali et al., WSDM 2023 | Provides the implementations of agenda-based and LLM-based user simulators, and metrics for the evaluation of conversational recommendation systems. |