HESML V1R4
HESML V1R4
HESML V1R4 is the fourth release of the Half-Edge Semantic Measures Library (HESML) [1], which is a new, scalable and efficient Java software library of ontology-based semantic similarity measures and Information Content (IC) models based on WordNet.
HESML V1R4 implements most ontology-based semantic similarity measures and Information Content (IC) models based on WordNet reported in the literature. In addition, it provides a XML-based input file format in order to specify the execution of reproducible experiments on WordNet-based similarity, even with no software coding.
HESML is introduced and detailed in a companion reproducibility paper [1] of the methods and experiments introduced in [2,3,4].
The main features of HESML are as follows: (1) it is based on an efficient and linearly scalable representation for taxonomies called PosetHERep introduced in [1], (2) its performance exhibits a linear scalability as regards the size of the taxonomy, and (3) it does not use any caching strategy of vertex sets.
Main novelties in HESML V1R4
Main novelties provided by HESML V1R4 are as follows:
(1) software implementation of a fast version of the Wu&Palmer [7] similarity measure defined by the formula sim(a,b) = 2*depth(LCS(a,b)) / (depth(a) + depth(b))
(2) software implementation of three new similarity measures based on the evaluation of pre-trained word embedding models in EMB, UKB(ppv) and Nasari file formats. Each row of the EMB file format contains a word vector with the raw coordinates of each word in a fixed dimension, whilst coordinates of word vectors in the UKB file format are defined by WordNet synsets, and finally Nasari vectors represent BabelNet synsets by a weighted and ranked set of other BabelNet synsets. In addition, Nasari provides an additional file with the BabelNet synsets corresponding to each word. Similarity measure based on the EMB and UKB files use the standard cosine function to compute the degree of similarity between words, whilst the similarity measure based on the Nasari files use the weighted overlap fucntion as detaield by the authors in [8]. HESML V1R4 release distribution is also published in Mendeley. HESML V1R4 is distributed with a dozen of pre-trained embedding models which cannot be included in Github because of their large size. However, you can download the 'WordEmbeddings.zip' file from our Dataverse repository cited below (e-cienciaDatos), then you should extract its content onto the 'HESML_Library\WordEmbeddings' directory.
Lastra-Díaz, Juan J.;Goikoetxea, Josu; Hadj Taieb, Mohamed; Agirre, Eneko;García-Serrano, Ana;Ben Aouicha, Mohamed, 2018, "Word similarity benchmarks of recent word embedding models and ontology-based semantic similarity measures", https://doi.org/10.21950/AQ1CVX, e-cienciaDatos,
(3) software implementation of the IC model proposed by Cai et al. (2017)[6]
(4) software implementation of a IC-based similarity measure proposed by Cai et al. (2017)[6]
Licensing information
HESML library is freely distributed for any non-commercial purpose under a CC By-NC-SA-4.0 license, subject to the citing of the main HESML paper [1] as attribution requirement. On other hand, the commercial use of the similarity measures introduced in [2], as well as part of the intrinsic IC models introduced in [3] and [4], is protected by a patent application [5]. In addition, any user of HESML must fulfill other licensing terms described in [1] related to other resources distributed with the library, such as WordNet and a dataset of corpus-based IC models, among others.
References:
[1] Lastra-DÃaz, J. J., GarcÃa-Serrano, A., Batet, M., Fernández, M. and Chirigati, F. (2017). HESML: a scalable ontology-based semantic similarity measures library with a set of reproducible experiments and a replication dataset. Information Systems 66, 97-118. http://dx.doi.org/10.1016/j.is.2017.02.002
[2] Lastra-DÃaz, J. J., & GarcÃa-Serrano, A. (2015). A novel family of IC-based similarity measures with a detailed experimental survey on WordNet. Engineering Applications of Artificial Intelligence Journal, 46, 140-153. http://dx.doi.org/10.1016/j.engappai.2015.09.006
[3] Lastra-DÃaz, J. J., & GarcÃa-Serrano, A. (2015). A new family of information content models with an experimental survey on WordNet. Knowledge-Based Systems, 89, 509-526. http://dx.doi.org/10.1016/j.knosys.2015.08.019
[4] Lastra-DÃaz, J. J., & GarcÃa-Serrano, A. (2016). A refinement of the well-founded Information Content models with a very detailed experimental survey on WordNet. Universidad Nacional de Educación a Distancia (UNED). http://e-spacio.uned.es/fez/view/bibliuned:DptoLSI-ETSI-Informes-Jlastra-refinement
[5] Lastra DÃaz, J. J., & GarcÃa Serrano, A. (2016). System and method for the indexing and retrieval of semantically annotated data using an ontology-based information retrieval model. United States Patent and Trademark Office (USPTO) Application, US2016/0179945 A1.
[6] Cai, Y., Zhang, Q., Lu, W., & Che, X. (2017). A hybrid approach for measuring semantic similarity based on IC-weighted path distance in WordNet. Journal of Intelligent Information Systems, 1–25.
[7] Wu, Z., & Palmer, M. (1994). Verbs Semantics and Lexical Selection. In Proceedings of the 32nd Annual Meeting on Association for Computational Linguistics (pp. 133–138). Stroudsburg, PA, USA: Association for Computational Linguistics.
[8] Camacho-Collados, J., Pilehvar, M. T., & Navigli, R. (2016). Nasari: Integrating explicit knowledge and corpus statistics for a multilingual representation of concepts and entities. Artificial Intelligence, 240, 36–64.
Steps to reproduce the library
HESML V1R3 is distributed as a Java class library (HESML-V1R3.jar) plus a test driver application (HESMLclient.jar), which have been developed using NetBeans 8.0.2 for Windows, although it has been also compiled and evaluated on Linux-based platforms using the corresponding NetBeans versions.
The HESML-V1R3.jar file is already included in the HESML_Library\HESML\dist folder of the HESML_Release_V1R3.zip distribution file. In order to compile HESML from its source files, you must follow the following steps:
(1) Download the full distribution of HESML V1R3.
(2) Install Java 8, Java SE Dev Kit 8 and NetBeans 8.0.2 or higher in your workstation.
(3) Launch NetBeans IDE and open the HESML and HESMLclient projects contained in the root folder. NetBeans automatically detects the presence of a nbproject subfolder with the project files.
(4) Select HESML and HESMLclient projects in the project treeview respectively. Then, invoke the "Clean and Build project (Shift + F11)" command in order to compile both projects.
In order to remain up to date on new HESML versions, as well as asking for technical support, we invite the readers to subscribe to the HESML forum by sending an email to the following address:
hesml+subscribe@googlegroups.com
Steps to use the library
You can use the HESMLclient program to run reproducible experiments or create your own client programs using the HESMLclient code as example. For more information, including a detailed description of how to run the reproducible experiments and extending the library, we refer the reader to the paper [1] above.