AffectiveSpace 2: Enabling Affective Intuition for Concept-Level Sentiment Analysis
Authors: Erik Cambria, Jie Fu, Federica Bisio, Soujanya Poria
AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | an evaluation section proposes experimental results for an opinion mining task; finally, the last section provides some concluding remarks. Experimental Results In order to evaluate the new analogical reasoning model, a comparison between Affective Space and Affective Space 2 has been performed both over a benchmark for affective common-sense knowledge (BACK) (Cambria and Hussain 2012b), for directly testing the affective analogical reasoning capabilities of the two models, and over a dataset of natural language opinions, for comparing how the two different configurations of Affective Space (SVD-built versus RP-built) perform within the more practical task of concept-level opinion mining. Both vector space models were built upon the new 50k × 120k Affect Net matrix. |
| Researcher Affiliation | Academia | Erik Cambria School of Computer Engineering NTU, Singapore cambria@ntu.edu.sg Jie Fu School of Computing NUS, Singapore jie.fu@comp.nus.edu.sg Federica Bisio DITEN University of Genoa, Italy federica.bisio@edu.unige.it Soujanya Poria School of Computer Engineering NTU, Singapore sporia@ntu.edu.sg |
| Pseudocode | Yes | Therefore, given that the distance between two points in the space is defined as D(ei, ej) = rPd s=1 e(s) i e(s) j 2, the adopted algorithm can be sum- marized as follows: 1. Each centroid ei Rd (i = 1, 2, ..., k) is set as one of the 24 basic emotions of the Hourglass model; 2. Assign each instance ej to a cluster ei if D(ej, ei) D(ej, ei ) where i(i ) = 1, 2, ..., k; 3. Find a new centroid ei for each cluster c so that P j Cluster c D(ej, ei) P j Cluster c D(ej, ei ); 4. Repeat step 2 and 3 until no changes on centroids are observed. |
| Open Source Code | No | No concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described in this paper was found. |
| Open Datasets | No | The paper uses "BACK" and "Patient Opinion" datasets, citing the source of the raw data (LiveJournal and patientopinion.org.uk) and publications related to their construction (Cambria and Hussain 2012b). However, it does not provide concrete access (link, DOI, specific repository, or formal citation to a publicly available processed version) to the exact datasets or specific processed splits used in their experiments. |
| Dataset Splits | No | The paper does not provide specific dataset split information (exact percentages, sample counts, or detailed splitting methodology) for training, validation, or testing. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper mentions general algorithms and methods like PCA, SVD, and random projection, but does not provide specific ancillary software details with version numbers (e.g., library or solver names with versions). |
| Experiment Setup | No | The paper describes the overall structure and components of the sentic computing engine and the clustering algorithm but does not provide specific experimental setup details such as hyperparameter values, training configurations, or optimizer settings. |