A Discriminative Latent Variable Model for Online Clustering

Authors: Rajhans Samdani, Kai-Wei Chang, Dan Roth

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In our experiments on coreference resolution and document clustering, L3M outperforms several existing online as well as batch supervised clustering techniques. We present experiments on coreference resolution and document clustering.
Researcher Affiliation Collaboration Rajhans Samdani, Google Research RAJHANS@GOOGLE.COM Kai-Wei Chang, University of Illinois KCHANG10@ILLINOIS.EDU Dan Roth, University of Illinois DANR@ILLINOIS.EDU
Pseudocode No The paper describes algorithms but does not include structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described in this paper.
Open Datasets Yes We show experimental results on two benchmark English coreference datasets ACE 2004 (NIST, 2004) and Ontonotes-5.0 (Pradhan et al., 2012).
Dataset Splits Yes ACE 2004 data contains 442 documents, split into 268 training, 68 development, and 106 testing documents... Onto Notes-5.0 (Pradhan et al., 2012) is the largest annotated corpus on coreference with a total of 3,145 training documents and 348 testing documents. We use 343 documents from the training set for validation.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions using structural SVMs and an ILP solver, but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes For all the algorithms, we tune the regularization parameters (and also γ for L3M) to optimize the targeted evaluation metric on the development set. For all the online clustering techniques (Sum-Link, Bin.-Left-Link, L3M), we present results with a single pass over the data as well as with multiple number of passes tuned on a validation set... For L3M (tuned γ), the best value of γ for ACE 2004 for one pass was 0; with multiple passes, the best γ was 0.2. ... it took five passes to acheive top performance on the development set for both the datasets and for all the online algorithms.