Abstract:Knowledge about protein-protein interactions (PPIs) unveils the molecular mechanisms of biological processes. The volume and content of published biomedical literature on protein interactions is expanding rapidly, making it increasingly difficult for interaction database administrators, responsible for content input and maintenance to detect and manually update protein interaction information. The objective of this work is to develop an effective approach to automatic extraction of PPI information from biomedical literature.We present a weighted multiple kernel learning-based approach for automatic PPI extraction from biomedical literature. The approach combines the following kernels: feature-based, tree, graph and part-of-speech (POS) path. In particular, we extend the shortest path-enclosed tree (SPT) and dependency path tree to capture richer contextual information.Our experimental results show that the combination of SPT and dependency path tree extensions contributes to the improvement of performance by almost 0.7 percentage units in F-score and 2 percentage units in area under the receiver operating characteristics curve (AUC). Combining two or more appropriately weighed individual will further improve the performance. Both on the individual corpus and cross-corpus evaluation our combined kernel can achieve state-of-the-art performance with respect to comparable evaluations, with 64.41% F-score and 88.46% AUC on the AImed corpus.As different kernels calculate the similarity between two sentences from different aspects. Our combined kernel can reduce the risk of missing important features. More specifically, we use a weighted linear combination of individual kernels instead of assigning the same weight to each individual kernel, thus allowing the introduction of each kernel to incrementally contribute to the performance improvement. In addition, SPT and dependency path tree extensions can improve the performance by including richer context information.

The Role of Dependency Information for Protein-Protein Interaction Extraction

Exploiting dependency information for feature-based protein-protein interaction extraction

Dependency-driven feature-based learning for extracting protein-protein interactions from biomedical text

Tree Kernel-Based Protein-Protein Interaction Extraction from Biomedical Literature.

Integrating Semantic Information into Multiple Kernels for Protein-Protein Interaction Extraction from Biomedical Literatures

Multiple kernel learning in protein-protein interaction extraction from biomedical literature.

Protein-protein interaction extraction from biomedical literatures based on a combined kernel

Protein-Protein Interaction Extraction Based on Ensemble Kernel

Automatic Extraction of Protein Interaction in Literature

Extracting protein-protein interaction from biomedical literature using an ensemble kernel

Neighborhood hash graph kernel for protein-protein interaction extraction.

Protein-Protein Interaction Extraction: A Supervised Learning Approach

A General Protein-Protein Interaction Extraction Architecture Based on Word Representation and Feature Selection.

BioPPISVMExtractor: a protein-protein interaction extractor for biomedical literature using SVM and rich feature sets.

Hash Subgraph Pairwise Kernel for Protein-Protein Interaction Extraction.

SVM-based Protein-Protein Interaction Extraction from Medline Abstracts

Ranking support vector machine for multiple kernels output combination in protein-protein interaction extraction from biomedical literature.

Protein-Protein Interaction Extraction: A Supervised Learning Approach}

Protein-Protein Interaction extraction based on ensemble kernel model and active learning strategy

A hybrid method for extraction of protein-protein interactions from literature

A Hybrid Protein-Protein Interaction Triple Extraction Method for Biomedical Literature.