Use Of Semantic Similarity Measures Research Articles

A Semi-Automated Approach to Process Discovery from Multiple Natural Language Documents

The discovery of process information in natural language documents is hindered by the innate ambiguity of natural language. This issue is even more challenging when process information is spread throughout several documents, each with the possibility of different formats, writers, and original purposes, that must be read and interpreted individually by a business analyst before their information can be uncovered. This work presents a semi-automated approach to process discovery from multiple documents simultaneously, making use of semantic similarity measures and natural language processing techniques to identify, gather, and extract process information from these documents. An experiment is conducted with this approach to demonstrate its use and results.

Reading development at the text level: an investigation of surprisal and embeddingbased text similarity effects on eyemovements in Chinese early readers.

This paper describes the use of semantic similarity measures based on distributed representations of words, sentences, and paragraphs (so-called “embeddings”) to assess the impact of supra-lexical factors on eye-movement data from early readers of Chinese. In addition, we used a corpus-based measure of surprisal to assess the impact of local word predictability. Eye movement data from 56 Chinese students were collected (a) in the students’ 4th grade and (b) one year later while they were in 5th grade. Results indicated that surprisal and some text similarity measures have a significant impact on the momentto- moment processing of words in reading. The paper presents an easy-to-use set of tools for linking the low-level aspects of fixation durations to a hierarchy of sentence-level and paragraph-level features that can be computed automatically. The study is the first attempt, as far as we are aware, to track the developmental trajectory of these influences in developing readers across a range of reading abilities. The similarity-based measures described here can be used (a) to provide a measure of reader sensitivity to sentence and paragraph cohesion and (b) to assess specific texts for their suitability for readers of different reading ability levels.

Ontology-driven integrative analysis of omics data through Onassis

Public repositories of large-scale omics datasets represent a valuable resource for researchers. In fact, data re-analysis can either answer novel questions or provide critical data able to complement in-house experiments. However, despite the development of standards for the compilation of metadata, the identification and organization of samples still constitutes a major bottleneck hampering data reuse. We introduce Onassis, an R package within the Bioconductor environment providing key functionalities of Natural Language Processing (NLP) tools. Leveraging biomedical ontologies, Onassis greatly simplifies the association of samples from large-scale repositories to their representation in terms of ontology-based annotations. Moreover, through the use of semantic similarity measures, Onassis hierarchically organizes the datasets of interest, thus supporting the semantically aware analysis of the corresponding omics data. In conclusion, Onassis leverages NLP techniques, biomedical ontologies, and the R statistical framework, to identify, relate, and analyze datasets from public repositories. The tool was tested on various large-scale datasets, including compendia of gene expression, histone marks, and DNA methylation, illustrating how it can facilitate the integrative analysis of various omics data.

Cross Domain Recommendation Using Semantic Similarity and Tensor Decomposition

Online shopping has become the buzzword in this information age. Users want to purchase the best possible item and services at the shortest span of time. In this information age Recommender system is a very useful tool, because it has the capability of filtering the information according to user interest and provide personalized suggestion. One of the major drawbacks of the classical recommender system is that, they deal with the only single domain. In real world scenario domains could be related to each other by some common information. There are many approaches available for cross domain recommendation, but they are not able to provide better accuracy of high dimensional data and these approaches are suffering from data sparsity problem. In this paper, we deal with cross domain recommendation where we exploit knowledge from auxiliary domains (e.g., movies) which contains additional user preference data to improve recommendation on the target domain (e.g., books). In order to achieve a high level of accuracy, we make use of semantic similarity measure of common information by which domains are related and Tensor decomposition to exploiting the latent factor for high dimensional data. Tensor decomposition with semantic similarity is used for making cross domain recommendation where in the data sparsity problem is avoided by normalizing and clustering the data in auxiliary domain. We provide experimental results on real world data sets and compared our proposed method with other similar approaches based on hit ratio and the results show that we achieve a better hit ratio.

The use of semantic similarity measures for optimally integrating heterogeneous Gene Ontology data from large scale annotation pipelines.

With the advancement of new high throughput sequencing technologies, there has been an increase in the number of genome sequencing projects worldwide, which has yielded complete genome sequences of human, animals and plants. Subsequently, several labs have focused on genome annotation, consisting of assigning functions to gene products, mostly using Gene Ontology (GO) terms. As a consequence, there is an increased heterogeneity in annotations across genomes due to different approaches used by different pipelines to infer these annotations and also due to the nature of the GO structure itself. This makes a curator's task difficult, even if they adhere to the established guidelines for assessing these protein annotations. Here we develop a genome-scale approach for integrating GO annotations from different pipelines using semantic similarity measures. We used this approach to identify inconsistencies and similarities in functional annotations between orthologs of human and Drosophila melanogaster, to assess the quality of GO annotations derived from InterPro2GO mappings compared to manually annotated GO annotations for the Drosophila melanogaster proteome from a FlyBase dataset and human, and to filter GO annotation data for these proteomes. Results obtained indicate that an efficient integration of GO annotations eliminates redundancy up to 27.08 and 22.32% in the Drosophila melanogaster and human GO annotation datasets, respectively. Furthermore, we identified lack of and missing annotations for some orthologs, and annotation mismatches between InterPro2GO and manual pipelines in these two proteomes, thus requiring further curation. This simplifies and facilitates tasks of curators in assessing protein annotations, reduces redundancy and eliminates inconsistencies in large annotation datasets for ease of comparative functional genomics.

Semantic Similarities as Discriminative Features of Protein Complexes

Biological data about genes, proteins and biologically relevant molecules that are stored in databases may be associated to biological information (knowledge) such as experiments, properties and functions, response to drugs etc. Such knowledge is formally structured into ontologies that provide the best formalize to organize and store knowledge. In the biological field, Gene Ontology (GO) provides both a categorization of annotating terms and a source of annotation for genes and proteins. Consequently it is possible to introduce novel methodologies of analysis that are based on the use of ontologies. Recently a growing interest has caputed semantic similarities, i.e. the calculation of the similarity of two or more proteins starting from their annotations. For instance semantic measures have been used for the prediction of protein complexes. Although the importance of these researches, some problems remain still unsolved: the assessment of semantic measures with respect to biological features as well as a deep study on the impact of the chosen measure in the obtained results. This paper focus on the use of semantic similarity measures into the protein complexes prediction pipeline. For these aims we investigated if there exists a bias among different measures as well as a higher value of semantic similarity within proteins that participate in the same complex. Results confirm that protein belonging to the same complex have a bigger average values of semantic similarity with respect to the average values of the proteomes. This confirm a possible use of semantic similarity measures within protein complexes prediction algorithms and a way to choose the best one among them. Keywords: Ontologies, protein interaction networks, semantic similarity measures.

Smart combination of web measures for solving semantic similarity problems

PurposeSemantic similarity measures are very important in many computer‐related fields. Previous works on applications such as data integration, query expansion, tag refactoring or text clustering have used some semantic similarity measures in the past. Despite the usefulness of semantic similarity measures in these applications, the problem of measuring the similarity between two text expressions remains a key challenge. This paper aims to address this issue.Design/methodology/approachIn this article, the authors propose an optimization environment to improve existing techniques that use the notion of co‐occurrence and the information available on the web to measure similarity between terms.FindingsThe experimental results using the Miller and Charles and Gracia and Mena benchmark datasets show that the proposed approach is able to outperform classic probabilistic web‐based algorithms by a wide margin.Originality/valueThis paper presents two main contributions. The authors propose a novel technique that beats classic probabilistic techniques for measuring semantic similarity between terms. This new technique consists of using not only a search engine for computing web page counts, but a smart combination of several popular web search engines. The approach is evaluated on the Miller and Charles and Gracia and Mena benchmark datasets and compared with existing probabilistic web extraction techniques.

Towards the assessment of semantic similarity analysis of protein data

Bioinformatics approaches to the study of proteins yield to the introduction of different methodologies and related tools for the analysis of different types of data related to proteins, ranging from primary, secondary and tertiary structures to interaction data [1], not to mention functional knowledge. One of the most advanced tools for encoding and representing functional knowledge in a formal way is the Gene Ontology (GO) [2,3]. It is composed of three ontologies, named Biological Process (BP), Molecular Function (MF) and Cellular Component (CC). Each ontology consists of a set of terms (GO terms) representing different functions, biological processes and cellular components within the cell. GO terms are connected each other to form a hierarchical graph. Terms representing similar functions are close to each other within this graph. Biological molecules are associated with GO terms that represent their functions, biological roles and localization. This process, usually referred to as annotation process, can be performed under the supervision of an expert or in a fully automated way. Obviously, computationally inferred annotations, commonly known as Electronically Inferred Annotation (IEA), are not as reliable as experimentally determined annotations. For this reason every annotation is labeled with an Evidence Code (EC) that keeps track of the type of process used to produce the annotation itself. Considering the release of annotations of April, 2010, about the 98% of all the annotations is an IEA annotation [4]. The term annotation corpus is commonly used to identify all the annotations involving a set of proteins or genes, usually referring the whole proteomes and genomes (i.e. the annotation corpus of yeast). For lack of space we do not further describe the Gene Ontology. A comprehensive review has been provided by du Plessis et al. [4] and by Guzzi et al. [5]. The availability of well formalized functional data enabled the use of computational methods to analyse genes and proteins from the functional point of view. For example, a set of algorithms, known as functional enrichment algorithms, have been developed to determine the statistical significance of the presence (or the absence) of a GO Term in a set of gene products. A detailed review of these algorithms can be found in [4]. An interesting problem is how to express quantitatively the relationships between GO terms. Several measures, referred to as (term) semantic similarity (SS) measures, has been introduced in the last decade. Given two or more GO terms, they try to quantify the similarity of the functional aspects represented by the terms within the cell. Exploiting annotation corpora, semantic similarity measures have been further extended to the evaluation of the similarity of genes and proteins on the basis of their annotations. Many different works have focused on the following tasks: (i) the definition of ad-hoc semantic similarity measures tailored to the characteristics of Gene Ontology; (ii) the definition of measures of comparison of genes and proteins; (iii) the introduction of methodologies for the systematic assessment of semantic similarity measures; (iv) the use of semantic similarity measures in many different contexts and applications. Despite its relevance, the application of semantic similarity for the systematic analysis of protein data is still an open research area. There are, in fact, two main questions that have to be addressed: (i) the systematic assessment of SS with respect to other biological features, i.e. how much an high or a low value of SS is biologically meaningful; (ii) how reliable are the SS themselves, i.e. is there any systematic error or bias in the calculation of SS? Both these problems are relevant for the diffusion of SS measures; while in the first case several approaches have been proposed, confronting SS measures with a pletora of different biological features, only few works dealt with the second problem in a systematic way [5,6,7].

A Web Search Engine-Based Approach to Measure Semantic Similarity between Words

Measuring the semantic similarity between words is an important component in various tasks on the web such as relation extraction, community mining, document clustering, and automatic metadata extraction. Despite the usefulness of semantic similarity measures in these applications, accurately measuring semantic similarity between two words (or entities) remains a challenging task. We propose an empirical method to estimate semantic similarity using page counts and text snippets retrieved from a web search engine for two words. Specifically, we define various word co-occurrence measures using page counts and integrate those with lexical patterns extracted from text snippets. To identify the numerous semantic relations that exist between two given words, we propose a novel pattern extraction algorithm and a pattern clustering algorithm. The optimal combination of page counts-based co-occurrence measures and lexical pattern clusters is learned using support vector machines. The proposed method outperforms various baselines and previously proposed web-based semantic similarity measures on three benchmark data sets showing a high correlation with human ratings. Moreover, the proposed method significantly improves the accuracy in a community mining task.