- Email: [email protected]
Automatic approach to enrich databases using ontology: Application in medical domain
Available online at www.sciencedirect.com
Available online at www.sciencedirect.com Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science (2017) 000–000 Procedia Computer Science 11200 (2017) 387–396 Procedia Computer Science 00 (2017) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
International Conference on Knowledge Based and Intelligent Information and Engineering International Conference Knowledge Based and Intelligent Information Systems, on KES2017, 6-8 September 2017, Marseille, Franceand Engineering Systems, KES2017, 6-8 September 2017, Marseille, France
Automatic Automatic approach approach to to enrich enrich databases databases using using ontology: ontology: Application Application in medical domain in medical domain a Universit´ e a Universit´ e
a* a Zina Zina Nakhla Nakhlaa*,, Kaouther Kaouther Nouira Nouiraa
de Tunis , BESTMOD Laboratory Institut Sup´erieur de Gestion, 41 Avenue de la Libert´e 2000 Bardo, Tunisia de Tunis , BESTMOD Laboratory Institut Sup´erieur de Gestion, 41 Avenue de la Libert´e 2000 Bardo, Tunisia
Abstract Abstract The enrichment of databases is fundamental to maintain them, as well as the consistency and accuracy of the data. The database The enrichment of databases is fundamental to maintain them, as well as the consistency and accuracy of the data. The database becomes useless if it is not up to date. Since there are a large number of databases, an automatic enrichment approach is required. becomes useless if it is not up to date. Since there are a large number of databases, an automatic enrichment approach is required. However, until now no efficient approach has been provided in order to cope with this problem. In this paper, we propose a new However, until now no efficient approach has been provided in order to cope with this problem. In this paper, we propose a new approach to automate the enrichment of databases. It is based on an ontology, which model domains through sets of concepts and approach to automate the enrichment of databases. It is based on an ontology, which model domains through sets of concepts and semantic relationships established between them. The proposed approach presents a set of rules to analyze ontologies and databases semantic relationships established between them. The proposed approach presents a set of rules to analyze ontologies and databases components and filter subsequently the necessary ones for the database enrichment of databases. We applied our approach in the components and filter subsequently the necessary ones for the database enrichment of databases. We applied our approach in the medical domain that is a renewable domain. Also, it is characterized by a large number of databases and ontologies, and a large medical domain that is a renewable domain. Also, it is characterized by a large number of databases and ontologies, and a large volume of data. For experimentations, a platform is developed to test rules using medical databases and medical ontologies. As a volume of data. For experimentations, a platform is developed to test rules using medical databases and medical ontologies. As a result we obtain enriched databases with new components that are either tables, attributes, or records. result we obtain enriched databases with new components that are either tables, attributes, or records. c 2017 The Authors. Published by Elsevier B.V. c 2017 2017 The The Authors.Published Published byElsevier ElsevierB.V. B.V. © Peer-review Authors. under responsibilityby of KES International. Peer-review under responsibility of KES International. International Keywords: Database; Ontology; Enrichment rules Keywords: Database; Ontology; Enrichment rules
1. Introduction 1. Introduction Database managment systems are used since 1970s to store various kinds of data for different purposes. They Database managment systems are used since 1970s to store various kinds of data for different purposes. They enabled information to be efficiently stored and queried. However, databases are not complete, the new elements of enabled information to be efficiently stored and queried. However, databases are not complete, the new elements of knowledge must be continually added 11 . On the other hand, ontologies have appeared as an alternative to databases knowledge must be continually added . On the other 2hand, ontologies have appeared as an alternative to databases in applications that require a more enriched meaning 2 . Several methods in the literature used ontology to model in applications that require a more enriched meaning . Several methods in the literature used ontology to model information systems in different domains 33 , to develop decision support systems 44 , to construct database based on support systems , to construct database based on information5 systems in different domains , to develop decision ontologies 5 , to improve the habitability of a natural language 66 , to analyze data from the patient record 77 ... Also, onontologies , to improve the habitability of a natural language , to analyze data from the patient record ... Also, on, tologies are widely used in different domains such as Semantic Web 88 , Natural Language Processing 99 , Medicine 10 tologies are11widely used in different domains such as Semantic Web , Natural Language Processing , Medicine 10 , Commerce 11 ,... An ontology can provide enough information about a domain, and even structure the appropriate Commerce ,... An ontology can provide enough information about a domain, and even structure the appropriate terms of a domain. Besides the similarity between ontologies and databases, ontologies provide summarized contexterms of a domain. Besides the similarity between ontologies and databases, ontologies provide summarized contex∗ ∗
Zina Nakhla. Tel.: +2-161-105-3411 ; fax: +0-000-000-0000. Zina Nakhla. Tel.: +2-161-105-3411 ; fax: +0-000-000-0000. E-mail address: [email protected] E-mail address: [email protected]
c 2017 The Authors. Published by Elsevier B.V. 1877-0509 c 2017 The Authors. Published by Elsevier B.V. 1877-0509 Peer-review under responsibility of KES International. 1877-0509 2017responsibility The Authors. Published by Elsevier B.V. Peer-review©under of KES International. Peer-review under responsibility of KES International 10.1016/j.procs.2017.08.221
388
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
tual information about the contents of databases. In this context, we propose to use ontologies for the enrichment of different components of databases. Researches in the literature have proposed several approaches to improve database structures and data using ontologies. An approach to build databases based on ontologies 12 , an approach to create ontologies based on databases 13 , an approach to use ontology to manage data and to make decisions 14 . In addition some researches proposed a semi-automatic approach based on ontologies to enrich databases 15 . Other works proposed approaches to enrich only a part of databases e.g tables, attributes, records 16 . The proposed works are limited, they only enrich one component of the database that is mostly attributes or records. But there is a lack of works concentrated on tables enrichment. Also, the approaches in the literature are semi-automatic which necessitate every time the intervention of an expert. Our proposed approach presents a set of enrichment rules which automate the databases enrichment using ontologies. The proposed enrichment rules treated all the components of the database (tables, attributes, records). We focus in this paper on medical domain, which is characterized by an important volume of scalable information. Our goal is to have an updated database compared to the original one. The new database contains the missing information, and the novelties of the medical domain which are related to the database. This paper is organized as follows: Section 2 presents works in the literature, which are related to using ontology to improve databases and methods to enrich databases. Section 3 explains the architecture of our proposed approach. Section 4 describes the enrichment rules. Section 5 presents the databases and ontologies used for the experiments and results. Finally, we give conclusion. 2. Related work In the literature, ontologies are used to improve databases in different levels. It is used to remove ambiguity in the structured query language of the database 17 . Also, ontologies are used to build and to model databases which take ontology as input and generates a database schema based on it 18,12,19 . In addition, ontologies based databases are used to categorize web pages 20 , and used to harmonize knowledge concepts in databases and models 21 ... For the database enrichment, researches used the records matching process. It consists on bringing together data from different databases about the same entity 22 . Matching records in databases is used in different domain, mainly medical domain because it has multiple legacy and information systems that support health care professionals 23 . Huang et al. 24 propose a Pathway And Gene Enrichment Database (PAGED) which is an online database that integrates gene-set-based prior knowledge as molecular patterns. The database resulted consists of disease-gene association data, curated and integrated from Online Mendelian Inheritance in Man database and the Genetic Association Database. Some approaches proposed to enrich databases using ontologies. Jesus et al. 16 proposed to extract semantic information from unstructured texts and transformed it in ontology. Then, ontology will be used to enrich the contents of textual attributes. Hamaz and Benchikha 15 proposed a reverse engineering process that aims to transform a relational database to an Ontology-Based Database. They attempt to add additional semantics behind applying the enrichment process. This enrichment process is semi-automated because that requires interactions of an expert. Yuan An et al. 25 developed the FormMapper system to enrich databases that accept user-created data entry forms, using ontologies and integrates them into existing databases in the same domain. The limitations of this approach are to include the missing correspondences and syntactic nature. Most of aforementioned approaches have proposed a solution to enrich database. But these works are semi-automatic and enrich only a part of databases. These works need each time the intervention of experts, which is waste of time and effort. Also, some errors can be unnoticed by experts which engender errors in databases. Our work proposes an automatic and complete database enrichment approach. 3. Our proposed database enrichment approach Usually we find different computer applications related to similar databases. Despite the different issues of applications, they use nearly similar databases which can contain the same tables, most of the same attributes and records. Each database is created separately, which is waste of time, effort and storage space. To resolve this issue, simply share the database to optimize the development of application. However, a richer description of the target database
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
389
semantics may be available in ontology form. We can investigate richer ontologies to create shared databases (see Fig. 1). Our approach based on ontology structure to determine the missing information in database.
Fig. 1. Motivation example of the proposed approach.
The main objective of our approach is to automate the enrichment of different components of a database. The proposed enrichment process is based on ontologies. Our approach focuses on data but also on the database structure. The general process followed to enrich a databases from ontologies is depicted in Fig. 2. The enrichment process started by selecting ontology components (concepts, objectproperties, dataproperties, instances) and database components (tables, attributes, records). Then, it compares the semantics of the selected components. The comparison is based on predetermined conditions, which are presented in a set of enrichment rules. Enrichment rules will be explained in the next section. After the semantic analysis, we obtain a text document describing the enrichment instructions that will be performed on the database. The final step of the enrichment process is to use this document for generating the enriched database.
Fig. 2. General architecture of the proposed approach.
4. Enrichment rules The enrichment rules of our proposed approach are divided into two categories. The first category consists on the enrichment of database records using ontology instances. The second category consists on the enrichment of database structure e.g table, attribute using ontology components e.g concepts, objectproperties, dataproperties. 4.1. Records enrichment Instances represents data in ontologies, so they can be used to enrich records in databases. In this category, we analyze the concepts and the tables, and we select the appropriate instances for records enrichment. The selection of
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
390
ontology instances must follow some conditions: If concept in the ontology and a table in the database are similar e.g having the same name or synonyms, then instances will be added as a record in the table (see Fig. 3). Also, instances of concept and records of table will be compared to avoid redundancy. Rule 1 describes the enrichment of the database by records. Rule 1: When C ∈ O and T ∈ DB and S imilarity(C, T )
T hen C = Redundancy(Instances(C), records(T ))
(1)
Add − Records(T, C ) Where: O: means ontology. Similarity(C, T): returns true if names of concept C and table T are similars or synonyms. Redundancy(Data1, Data2): compares between two lists of data, and returns data whithout redundancy. Add-Records(T, I): Adds instances I as records in table T. Instances(C): returns instances of concept C. Records(T): returns records of table T.
Fig. 3. Enrichment of database records using one concept.
In the case, when two concepts in ontology can be used to enrich records of one table in database. We compare the similarity between concepts and table, and we compare instances of the two concepts and records to avoid redundancy (see Fig. 4). Rule 2 describes this case. Rule 2: When C1, C2 ∈ O and T ∈ DB and S imilarity(C1, T ) and S imilarity(C2, T )
T hen I = Merge(Instances(C1), Instances(C2)), C = (Redundancy(I, records(T ))
(2)
Add − Records(T, C ). Where: Merge(Data1, Data2): merges two lists of instances, and returns one list. The enrichment of records is based on an assumption that the attributes describing an instance of concept are the same as those in table. If the attributes are different, we use rule 5, presented in the next section, to ajust attributes.
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
391
Fig. 4. Enrichment of records database using two concepts.
4.2. Enrichment of structure This rules category deals with the databases structure enrichment such as adding table and attributes to database. If C1 and C2 two concepts in ontology linked by a relationships. If C1 is represented in the database by the table T and C2 not appear in a database then we can add C2 as a table (see Fig5.b). Rule 3 describes this case. Rule 3: When C1, C2 ∈ O and T ∈ DB and S imilarity(C1, T )
and Find − S imilarity(C1, T able(DB)) and Relation(C1, C2)
(3)
T hen Add − T able(C2, DB). Where: Add-Table(C, DB): adds concept C as table to DB and defines on its properties as primary key. Table(DB): returns list of tables of a database DB. Find-Similarity(C, Tables): returns true if C is similar to one of tables in database. Relation(C1, C2): returns true if relation exists between two concepts. If two concepts in the ontology are linked and these concepts appear as tables in the database. These tables have not link between them using primary-key and foreign-key. Based on the relation of concepts, we can complete the missing relation. We migrate primary-key of table as foreign-key in the second table (see Fig. 5(a)). Rule 4 describes this case. Rule 4: When C1, C2 ∈ O and T 1, T 2 ∈ DB and S imilarity(C1, T 1) and S imilarity(C2, T 2) and Relation(C1, C2) and ¬Related − tables(T 1, T 2)
(4)
T hen Add − Relation(T 1, T 2). Where: Related-tables(T1, T2): returns true if two tables are linked with primary-key and foreign-key. Add-Relation(T1, T2): Migates primary key of table T1 as a foreign key in table T2. We can enrich table of database with attributes. If a concept in the ontology appears as a table in the database. We compare between dataproperties of a concept and attributes of table. The dataproperty does not appear as attribute, will be added as attribute in the relative table (see Fig. 5(c)). In this case we compare and adjust the table column and dataproperties of a concept. Rule 5 describes this case.
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
392
Rule 5: When C ∈ O and T ∈ DB and S imilarity(C, T )
T hen P = UnS imilar − Prop(DataProperties(C), Attributes(T ))
(5)
Add − Attributes(T, P) Where: Add-Attributes(T, C): adds properties of concept C as attributes in table T. DataProperties(C): returns dataproperties of a concept C. Attributes(T): returns attributes of a table T. UnSimilar-Prop(P, A): returns DataProperties P of a concept different of Attributes A of table.
Fig. 5. Enrichment the structure of database.
5. Experiments and results For the experiments, we choose medical domain because it is always renewable and it is characterized by a large volume of data. Also, medical domain have a large number of databases and ontologies. We used five different medical databases to test our enrichment rules, and the selection is based on the variation of specialties of each database. Table 1 shows the number of different components of the selected databases. We use for the enrichment two different medial ontologies: (1) ’Symptom ontology’1 : designed around the guiding concept of a symptom and contains more than 900 concepts, it was developed by the Institute for Genome Sciences (IGS) at the University of Maryland. 1
https://bioportal.bioontology.org/ontologies/SYMP
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
393
(2)’Cardiovascular disease ontology’2 : designed to describe entities related to cardiovascular diseases (including the diseases themselves, the underlying disorders, and the related pathological processes). It is being developed at Sherbrooke University (Canada) and the Institut National de la Sant´e et de la Recherche M´edicale (INSERM). Both of ontology are coded with Ontology Web Langage (OWL). To implement our platform we used Eclipse and OWL API. We used SNOMED-CT ontology 3 to substruct similarity between concepts, which is the most comprehensive and precise clinical health terminology product in the world. Table 1. Medical databases used for the experimentation.
Database DB1: Disease Database DB2: Cardiovasculair Database DB3: Allergy Database DB4: Surgical Database DB5: Laboratory analysis Database
Tables 10 11 15 9 12
Records 43 21 53 28 27
Attributes 32 43 63 30 49
The performance of our approach is assessed by the evaluation of generated database: (1) quantitative evaluation and (2) qualitative evaluation. 5.1. Quantitative evaluation We evaluate the quantity of the new information stored in the database. We have computed the enrichment rate of the database, which is the percentage of the number of database components changement after enrichment by comparing to number of components before the enrichment. The number of components consists on tables, records, attributes, relationships between tables using primary and foreign keys. E1 computed the sum of components of database after the enrichment and. E2 presents the number of changements before the applying of the enrichment rules. (E1 − E2 )/E2 ∗ 100
(6)
Fig. 6 illustrates the enrichment rates for each databases as illustrated in equation on (6). Table 2 resumes the enriched results obtained using Symptom ontology (O1 ) and Cardiovascular ontology (O2 ). According to the sketched histograms, we can point out that most of enrichment rates of databases using O1 are higher than O2 . We remark that DB4 has the highest rate for the two ontologies, which explained by the similarity between the components of the ontology and DB4 . It is a surgical database, which is related to the topics of the two used ontologies. Besides DB4 have a small number of tables which proves the enrichment need. In contrast, DB2 have a small rater of changement when we use O2 . The explanation is that DB2 and O2 both are specialized on cardiovascular domain, so it contains most of similar components and does not require a high rate of change. We conclude that the enrichment of databases is based on the relation, between databases and ontologies, and the completeness of databases. 5.2. Qualitative evaluation The qualitative evaluation consists on evaluating the information stored in database. Experts evaluate the resultant database based on five quality criteria 26 : (1) Accessibility: the ability to access to database, (2) Accuracy: the degree to which the data mirrors the characteristics of the real world object or objects it represents, (3) Timely: the time the real world event being recorded occurred, (4) Completeness: the absence of blank and missed values, and (5) Consistency: the absence of difference, when comparing two or more representations of a thing against a definition. We asked twenty experts to judge the 2 3
http://www.obofoundry.org/ontology/cvdo.html http://www.snomed.org/snomed-ct
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
394
Table 2. The enrichment rates for databases using O1 and O2.
Database
E1
DB1 DB2 DB3 DB4 DB5
93 84 145 75 98
E2 103 101 172 104 107
O1
(%) 10.75 20.23 18.36 38.66 9.189
E2 97 86 164 113 109
O2
(%) 4,12 2,32 11,58 33,62 10,09
Fig. 6. Enrichment rates.
information quality of new databases compared to the last version of database. To judge we have two dimensions of satisfaction (high and low). Table 3 presents the number of expert for each dimension and for each criteria. Based on table 3, we conclude that the number of experts, which are satisfied of resulted databases, is higher than the number of experts which ae not satisfied of the quality of databases. Criteria that are more satisfactory for users are accessibity, completeness, and consistency. It proves that users have not problems in the access to databases. Also, it proves that data in databases are correct, complete, and characterized by the absence of ambiguity. Timely criterion is satisfactory for users. However, accuracy criterion is unsatisfactory for some users. This criterion depends to users because accuracy is the closeness of terms used in the database to the terms accepted by users. Furthermore, the selected databases belong to medical domain which characterized by several terms. Accuracy is correlated with context which explains the diversity in the level of satisfaction of users for accuracy criterion. Consequently, terms in database are considered accurate for some users and inaccurate for others. The results of qualitative evaluation show that the enriched databases are satisfying in term of information quality mainly for criteria of accessibility, completeness, and consistency. 5.3. Discussion We can summarize the performance of our approach based on above evaluations of experiments. Both quantitative and qualitative evaluations highlights the efficiency of our automatic approach of database enrichment. Table 2 presents the results of quantitative evaluation which are the rate of enrichment for each database. The obtained results show the performance of our approach for different databases. Then, we evaluate the quality of database after the enrichment. Table 3 presents the level of satisfaction of experts based on five quality criteria. This table shows that
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
Table 3. Level of satisfaction of the experts according to quality criteria for each database.
Information quality criteria
Accessibility Accuracy Timely Completeness Consistency Total
DB1
High 20 10 12 19 15 76
DB2
Low 00 10 08 01 05 24
High 18 12 14 16 14 74
DB3
Low 02 08 06 04 06 26
High 15 11 13 18 17 74
Low 05 09 07 02 03 26
DB4 High 20 09 12 15 17 73
395
DB5 Low 00 11 08 05 03 27
High 18 14 15 17 13 77
Low 02 06 05 03 07 23
experts are satisfying with the enriched database in the most of criteria. Through all the evaluation of experiments, we demonstrate that our approach has enhanced the enrichment of database significantly. 6. Conclusion We proposed an approach to automate the enrichment of database using ontology. The goal was to achieve an automatic way to analyze and compare between the components of ontology and database. We presented in this paper a set of enrichment rules which are divided in two categories: the first category is for the database enrichment using records and the second category is for the enrichment database structure enrichment. We developed a platform to apply rules and we use five medical databases and two medical ontologies. We evaluate the resulted databases with quantitative and qualitative evaluations. The quantitative evaluation shows that the number of database component grows after the application of rules. The qualitative evaluation shows that resulted database meets the quality criteria. As a future work, we will test our approach using voluminous databases and ontologies for different domain. And we will deal with the addition of new semantic relations using the inference rules. References 1. Dobrescu, E.M., PociovliÅteanu, D.M.. Some observations regarding the international databases. Procedia Economics and Finance 2015; 32:841 – 849. 2. Martinez-Cruz, C., Blanco, I.J., Vila, M.A.. Ontologies versus relational databases: Are they so different? a comparison. Artificial Intelligence Review 2011;38(4):271–290. 3. Viale, P., Bora, J.J., Benegui, M., Basualdo, M.. Human endocrine system modeling based on ontologies. Knowledge-Based Systems 2016; 111:113 – 132. 4. Kontopoulos, E., Martinopoulos, G., Lazarou, D., Bassiliades, N.. An ontology-based decision support tool for optimizing domestic solar hot water system selection. Journal of Cleaner Production 2016;112, Part 5:4636 – 4646. 5. Nakhla, Z., Nouira, K.. Ontology based database for computer medical system. In: 2nd Science And Information conference (SAI), IEEE. IEEE; 2014, p. 443 – 446. 6. Damljanovi, D., Agatonovi, M., Cunningham, H., Bontcheva, K.. Improving habitability of natural language interfaces for querying ontologies with feedback and clarification dialogues. Web Semantics: Science, Services and Agents on the World Wide Web 2013;19:1 – 21. 7. Hsu, W., Gonzalez, N.R., Chien, A., Villablanca, J.P., Pajukanta, P., Viuela, F., et al. An integrated, ontology-driven approach to constructing observational databases for research. Journal of Biomedical Informatics 2015;55:132 – 142. 8. Ding, L., Kolari, P., Ding, Z., Avancha, S.. Using Ontologies in the Semantic Web: A Survey. Boston, MA: Springer US; 2007, p. 79–113. 9. Estival, D., Nowak, C., Zschorn, A.. Towards ontology-based natural language processing. In: Proceeedings of the Workshop on NLP and XML (NLPXML-2004): RDF/RDFS and OWL in Language Technology; NLPXML ’04. Stroudsburg, PA, USA: Association for Computational Linguistics; 2004, p. 59–66. 10. Jovic, A., Prcela, M., Gamberger, D.. Ontologies in medical knowledge representation. In: Proceeding of the 29th International Conference on. Information Technology Interfaces, ( ITI’07). IEEE; 2007, p. 535–540. 11. Li, G., Peng, Q.. Using ontology and rough set building e-commerce knowledge management system. In: Proceeding of Advances in Computer Science, Intelligent System and Environment (CSISE 2011); vol. 3. 2011, p. 13–18. 12. Nakhla, Z., Nouira, K.. Ontology maps to database: Ontomapdb approach. In: International Conference on Control, Engineering Information Technology (CEIT’13). IEEE; 2013, p. 103–108. 13. Gali, A., Chen, C.X., Claypool, K.T., Uceda-Sosa, R.. From Ontology to Relational Databases. Berlin, Heidelberg: Springer Berlin Heidelberg; 2004, p. 278–289.
396
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
14. Chuprina, S., Postanogov, I., Nasraoui, O.. Ontology based data access methods to teach students to transform traditional information systems and simplify decision making process. Procedia Computer Science 2016;80:1801 – 1811. 15. Kamal, H., Fouzia, B.. Semantic enrichment of relational databases. In: Second International Workshop on Advanced Information Systems for Enterprises (IWAISE). 2012, p. 15–19. 16. Campa˜na, J.R., Mart´ın-Bautista, M.J., Medina, J.M., Yager, R.R., Bulskov, H., Christiansen, H., et al. Semantic Enrichment of Database Textual Attributes. Berlin, Heidelberg: Springer Berlin Heidelberg; 2009, p. 488–499. 17. Sujatha, B., Raju, S.V.. Ontology based natural language interface for relational databases. Procedia Computer Science 2016;92:487 – 492. 2nd International Conference on Intelligent Computing, Communication amp; Convergence, {ICCC} 2016, 24-25 January 2016, Bhubaneswar, Odisha, India. 18. LePendu, P., Dou, D.. Using ontology databases for scalable query answering, inconsistency detection, and data integration. J Intell Inf Syst 2011;37(2):217–244. 19. Sun, Z., Finkelstein, A., Ashmore, J.. Using Ontology with Semantic Web Services to Support Modeling in Systems Biology. Berlin, Heidelberg: Springer Berlin Heidelberg; 2007, p. 41–51. 20. Bai, R., Wang, X., Liao, J.. Using an Integrated Ontology Database to Categorize Web Pages. Berlin, Heidelberg: Springer Berlin Heidelberg; 2010, p. 300–309. 21. Wien, J., Knapen, M., Janssen, S., Verweij, P., Athanasiadis, I., Li, H., et al. Using ontology to harmonize knowledge concepts in data and models. 2007. 22. Verykios, V.. Automating the approximate record-matching process. Information Sciences 2000;126(1-4):83–98. 23. Wang, X., Ling, J.. Multiple valued logic approach for matching patient records in multiple databases. Journal of Biomedical Informatics 2012;45(2):224 – 230. 24. Huang, H., Wu, X., Sonachalam, M., Mandape, S.N., Pandey, R., MacDorman, K.F., et al. Paged: a pathway and gene-set enrichment database to enable molecular phenotype discoveries. BMC bioinformatics 2012;13(Suppl 15):S2. 25. An, Y., Khare, R., Song, I.Y., Hu, X.. Automatically Mapping and Integrating Multiple Data Entry Forms into a Database. In: Jeusfeld, M., Delcambre, L., Ling, T.W., editors. Conceptual Modeling ER 2011; vol. 6998 of Lecture Notes in Computer Science. Springer Berlin Heidelberg; 2011, p. 261–274. 26. Fehrenbacher, D.D.. Perceptions of information quality dimensions from the perspective of commodity theory. Behavior and Information Technology 2016;35(4):254–267.
Available online at www.sciencedirect.com Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science (2017) 000–000 Procedia Computer Science 11200 (2017) 387–396 Procedia Computer Science 00 (2017) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
International Conference on Knowledge Based and Intelligent Information and Engineering International Conference Knowledge Based and Intelligent Information Systems, on KES2017, 6-8 September 2017, Marseille, Franceand Engineering Systems, KES2017, 6-8 September 2017, Marseille, France
Automatic Automatic approach approach to to enrich enrich databases databases using using ontology: ontology: Application Application in medical domain in medical domain a Universit´ e a Universit´ e
a* a Zina Zina Nakhla Nakhlaa*,, Kaouther Kaouther Nouira Nouiraa
de Tunis , BESTMOD Laboratory Institut Sup´erieur de Gestion, 41 Avenue de la Libert´e 2000 Bardo, Tunisia de Tunis , BESTMOD Laboratory Institut Sup´erieur de Gestion, 41 Avenue de la Libert´e 2000 Bardo, Tunisia
Abstract Abstract The enrichment of databases is fundamental to maintain them, as well as the consistency and accuracy of the data. The database The enrichment of databases is fundamental to maintain them, as well as the consistency and accuracy of the data. The database becomes useless if it is not up to date. Since there are a large number of databases, an automatic enrichment approach is required. becomes useless if it is not up to date. Since there are a large number of databases, an automatic enrichment approach is required. However, until now no efficient approach has been provided in order to cope with this problem. In this paper, we propose a new However, until now no efficient approach has been provided in order to cope with this problem. In this paper, we propose a new approach to automate the enrichment of databases. It is based on an ontology, which model domains through sets of concepts and approach to automate the enrichment of databases. It is based on an ontology, which model domains through sets of concepts and semantic relationships established between them. The proposed approach presents a set of rules to analyze ontologies and databases semantic relationships established between them. The proposed approach presents a set of rules to analyze ontologies and databases components and filter subsequently the necessary ones for the database enrichment of databases. We applied our approach in the components and filter subsequently the necessary ones for the database enrichment of databases. We applied our approach in the medical domain that is a renewable domain. Also, it is characterized by a large number of databases and ontologies, and a large medical domain that is a renewable domain. Also, it is characterized by a large number of databases and ontologies, and a large volume of data. For experimentations, a platform is developed to test rules using medical databases and medical ontologies. As a volume of data. For experimentations, a platform is developed to test rules using medical databases and medical ontologies. As a result we obtain enriched databases with new components that are either tables, attributes, or records. result we obtain enriched databases with new components that are either tables, attributes, or records. c 2017 The Authors. Published by Elsevier B.V. c 2017 2017 The The Authors.Published Published byElsevier ElsevierB.V. B.V. © Peer-review Authors. under responsibilityby of KES International. Peer-review under responsibility of KES International. International Keywords: Database; Ontology; Enrichment rules Keywords: Database; Ontology; Enrichment rules
1. Introduction 1. Introduction Database managment systems are used since 1970s to store various kinds of data for different purposes. They Database managment systems are used since 1970s to store various kinds of data for different purposes. They enabled information to be efficiently stored and queried. However, databases are not complete, the new elements of enabled information to be efficiently stored and queried. However, databases are not complete, the new elements of knowledge must be continually added 11 . On the other hand, ontologies have appeared as an alternative to databases knowledge must be continually added . On the other 2hand, ontologies have appeared as an alternative to databases in applications that require a more enriched meaning 2 . Several methods in the literature used ontology to model in applications that require a more enriched meaning . Several methods in the literature used ontology to model information systems in different domains 33 , to develop decision support systems 44 , to construct database based on support systems , to construct database based on information5 systems in different domains , to develop decision ontologies 5 , to improve the habitability of a natural language 66 , to analyze data from the patient record 77 ... Also, onontologies , to improve the habitability of a natural language , to analyze data from the patient record ... Also, on, tologies are widely used in different domains such as Semantic Web 88 , Natural Language Processing 99 , Medicine 10 tologies are11widely used in different domains such as Semantic Web , Natural Language Processing , Medicine 10 , Commerce 11 ,... An ontology can provide enough information about a domain, and even structure the appropriate Commerce ,... An ontology can provide enough information about a domain, and even structure the appropriate terms of a domain. Besides the similarity between ontologies and databases, ontologies provide summarized contexterms of a domain. Besides the similarity between ontologies and databases, ontologies provide summarized contex∗ ∗
Zina Nakhla. Tel.: +2-161-105-3411 ; fax: +0-000-000-0000. Zina Nakhla. Tel.: +2-161-105-3411 ; fax: +0-000-000-0000. E-mail address: [email protected] E-mail address: [email protected]
c 2017 The Authors. Published by Elsevier B.V. 1877-0509 c 2017 The Authors. Published by Elsevier B.V. 1877-0509 Peer-review under responsibility of KES International. 1877-0509 2017responsibility The Authors. Published by Elsevier B.V. Peer-review©under of KES International. Peer-review under responsibility of KES International 10.1016/j.procs.2017.08.221
388
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
tual information about the contents of databases. In this context, we propose to use ontologies for the enrichment of different components of databases. Researches in the literature have proposed several approaches to improve database structures and data using ontologies. An approach to build databases based on ontologies 12 , an approach to create ontologies based on databases 13 , an approach to use ontology to manage data and to make decisions 14 . In addition some researches proposed a semi-automatic approach based on ontologies to enrich databases 15 . Other works proposed approaches to enrich only a part of databases e.g tables, attributes, records 16 . The proposed works are limited, they only enrich one component of the database that is mostly attributes or records. But there is a lack of works concentrated on tables enrichment. Also, the approaches in the literature are semi-automatic which necessitate every time the intervention of an expert. Our proposed approach presents a set of enrichment rules which automate the databases enrichment using ontologies. The proposed enrichment rules treated all the components of the database (tables, attributes, records). We focus in this paper on medical domain, which is characterized by an important volume of scalable information. Our goal is to have an updated database compared to the original one. The new database contains the missing information, and the novelties of the medical domain which are related to the database. This paper is organized as follows: Section 2 presents works in the literature, which are related to using ontology to improve databases and methods to enrich databases. Section 3 explains the architecture of our proposed approach. Section 4 describes the enrichment rules. Section 5 presents the databases and ontologies used for the experiments and results. Finally, we give conclusion. 2. Related work In the literature, ontologies are used to improve databases in different levels. It is used to remove ambiguity in the structured query language of the database 17 . Also, ontologies are used to build and to model databases which take ontology as input and generates a database schema based on it 18,12,19 . In addition, ontologies based databases are used to categorize web pages 20 , and used to harmonize knowledge concepts in databases and models 21 ... For the database enrichment, researches used the records matching process. It consists on bringing together data from different databases about the same entity 22 . Matching records in databases is used in different domain, mainly medical domain because it has multiple legacy and information systems that support health care professionals 23 . Huang et al. 24 propose a Pathway And Gene Enrichment Database (PAGED) which is an online database that integrates gene-set-based prior knowledge as molecular patterns. The database resulted consists of disease-gene association data, curated and integrated from Online Mendelian Inheritance in Man database and the Genetic Association Database. Some approaches proposed to enrich databases using ontologies. Jesus et al. 16 proposed to extract semantic information from unstructured texts and transformed it in ontology. Then, ontology will be used to enrich the contents of textual attributes. Hamaz and Benchikha 15 proposed a reverse engineering process that aims to transform a relational database to an Ontology-Based Database. They attempt to add additional semantics behind applying the enrichment process. This enrichment process is semi-automated because that requires interactions of an expert. Yuan An et al. 25 developed the FormMapper system to enrich databases that accept user-created data entry forms, using ontologies and integrates them into existing databases in the same domain. The limitations of this approach are to include the missing correspondences and syntactic nature. Most of aforementioned approaches have proposed a solution to enrich database. But these works are semi-automatic and enrich only a part of databases. These works need each time the intervention of experts, which is waste of time and effort. Also, some errors can be unnoticed by experts which engender errors in databases. Our work proposes an automatic and complete database enrichment approach. 3. Our proposed database enrichment approach Usually we find different computer applications related to similar databases. Despite the different issues of applications, they use nearly similar databases which can contain the same tables, most of the same attributes and records. Each database is created separately, which is waste of time, effort and storage space. To resolve this issue, simply share the database to optimize the development of application. However, a richer description of the target database
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
389
semantics may be available in ontology form. We can investigate richer ontologies to create shared databases (see Fig. 1). Our approach based on ontology structure to determine the missing information in database.
Fig. 1. Motivation example of the proposed approach.
The main objective of our approach is to automate the enrichment of different components of a database. The proposed enrichment process is based on ontologies. Our approach focuses on data but also on the database structure. The general process followed to enrich a databases from ontologies is depicted in Fig. 2. The enrichment process started by selecting ontology components (concepts, objectproperties, dataproperties, instances) and database components (tables, attributes, records). Then, it compares the semantics of the selected components. The comparison is based on predetermined conditions, which are presented in a set of enrichment rules. Enrichment rules will be explained in the next section. After the semantic analysis, we obtain a text document describing the enrichment instructions that will be performed on the database. The final step of the enrichment process is to use this document for generating the enriched database.
Fig. 2. General architecture of the proposed approach.
4. Enrichment rules The enrichment rules of our proposed approach are divided into two categories. The first category consists on the enrichment of database records using ontology instances. The second category consists on the enrichment of database structure e.g table, attribute using ontology components e.g concepts, objectproperties, dataproperties. 4.1. Records enrichment Instances represents data in ontologies, so they can be used to enrich records in databases. In this category, we analyze the concepts and the tables, and we select the appropriate instances for records enrichment. The selection of
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
390
ontology instances must follow some conditions: If concept in the ontology and a table in the database are similar e.g having the same name or synonyms, then instances will be added as a record in the table (see Fig. 3). Also, instances of concept and records of table will be compared to avoid redundancy. Rule 1 describes the enrichment of the database by records. Rule 1: When C ∈ O and T ∈ DB and S imilarity(C, T )
T hen C = Redundancy(Instances(C), records(T ))
(1)
Add − Records(T, C ) Where: O: means ontology. Similarity(C, T): returns true if names of concept C and table T are similars or synonyms. Redundancy(Data1, Data2): compares between two lists of data, and returns data whithout redundancy. Add-Records(T, I): Adds instances I as records in table T. Instances(C): returns instances of concept C. Records(T): returns records of table T.
Fig. 3. Enrichment of database records using one concept.
In the case, when two concepts in ontology can be used to enrich records of one table in database. We compare the similarity between concepts and table, and we compare instances of the two concepts and records to avoid redundancy (see Fig. 4). Rule 2 describes this case. Rule 2: When C1, C2 ∈ O and T ∈ DB and S imilarity(C1, T ) and S imilarity(C2, T )
T hen I = Merge(Instances(C1), Instances(C2)), C = (Redundancy(I, records(T ))
(2)
Add − Records(T, C ). Where: Merge(Data1, Data2): merges two lists of instances, and returns one list. The enrichment of records is based on an assumption that the attributes describing an instance of concept are the same as those in table. If the attributes are different, we use rule 5, presented in the next section, to ajust attributes.
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
391
Fig. 4. Enrichment of records database using two concepts.
4.2. Enrichment of structure This rules category deals with the databases structure enrichment such as adding table and attributes to database. If C1 and C2 two concepts in ontology linked by a relationships. If C1 is represented in the database by the table T and C2 not appear in a database then we can add C2 as a table (see Fig5.b). Rule 3 describes this case. Rule 3: When C1, C2 ∈ O and T ∈ DB and S imilarity(C1, T )
and Find − S imilarity(C1, T able(DB)) and Relation(C1, C2)
(3)
T hen Add − T able(C2, DB). Where: Add-Table(C, DB): adds concept C as table to DB and defines on its properties as primary key. Table(DB): returns list of tables of a database DB. Find-Similarity(C, Tables): returns true if C is similar to one of tables in database. Relation(C1, C2): returns true if relation exists between two concepts. If two concepts in the ontology are linked and these concepts appear as tables in the database. These tables have not link between them using primary-key and foreign-key. Based on the relation of concepts, we can complete the missing relation. We migrate primary-key of table as foreign-key in the second table (see Fig. 5(a)). Rule 4 describes this case. Rule 4: When C1, C2 ∈ O and T 1, T 2 ∈ DB and S imilarity(C1, T 1) and S imilarity(C2, T 2) and Relation(C1, C2) and ¬Related − tables(T 1, T 2)
(4)
T hen Add − Relation(T 1, T 2). Where: Related-tables(T1, T2): returns true if two tables are linked with primary-key and foreign-key. Add-Relation(T1, T2): Migates primary key of table T1 as a foreign key in table T2. We can enrich table of database with attributes. If a concept in the ontology appears as a table in the database. We compare between dataproperties of a concept and attributes of table. The dataproperty does not appear as attribute, will be added as attribute in the relative table (see Fig. 5(c)). In this case we compare and adjust the table column and dataproperties of a concept. Rule 5 describes this case.
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
392
Rule 5: When C ∈ O and T ∈ DB and S imilarity(C, T )
T hen P = UnS imilar − Prop(DataProperties(C), Attributes(T ))
(5)
Add − Attributes(T, P) Where: Add-Attributes(T, C): adds properties of concept C as attributes in table T. DataProperties(C): returns dataproperties of a concept C. Attributes(T): returns attributes of a table T. UnSimilar-Prop(P, A): returns DataProperties P of a concept different of Attributes A of table.
Fig. 5. Enrichment the structure of database.
5. Experiments and results For the experiments, we choose medical domain because it is always renewable and it is characterized by a large volume of data. Also, medical domain have a large number of databases and ontologies. We used five different medical databases to test our enrichment rules, and the selection is based on the variation of specialties of each database. Table 1 shows the number of different components of the selected databases. We use for the enrichment two different medial ontologies: (1) ’Symptom ontology’1 : designed around the guiding concept of a symptom and contains more than 900 concepts, it was developed by the Institute for Genome Sciences (IGS) at the University of Maryland. 1
https://bioportal.bioontology.org/ontologies/SYMP
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
393
(2)’Cardiovascular disease ontology’2 : designed to describe entities related to cardiovascular diseases (including the diseases themselves, the underlying disorders, and the related pathological processes). It is being developed at Sherbrooke University (Canada) and the Institut National de la Sant´e et de la Recherche M´edicale (INSERM). Both of ontology are coded with Ontology Web Langage (OWL). To implement our platform we used Eclipse and OWL API. We used SNOMED-CT ontology 3 to substruct similarity between concepts, which is the most comprehensive and precise clinical health terminology product in the world. Table 1. Medical databases used for the experimentation.
Database DB1: Disease Database DB2: Cardiovasculair Database DB3: Allergy Database DB4: Surgical Database DB5: Laboratory analysis Database
Tables 10 11 15 9 12
Records 43 21 53 28 27
Attributes 32 43 63 30 49
The performance of our approach is assessed by the evaluation of generated database: (1) quantitative evaluation and (2) qualitative evaluation. 5.1. Quantitative evaluation We evaluate the quantity of the new information stored in the database. We have computed the enrichment rate of the database, which is the percentage of the number of database components changement after enrichment by comparing to number of components before the enrichment. The number of components consists on tables, records, attributes, relationships between tables using primary and foreign keys. E1 computed the sum of components of database after the enrichment and. E2 presents the number of changements before the applying of the enrichment rules. (E1 − E2 )/E2 ∗ 100
(6)
Fig. 6 illustrates the enrichment rates for each databases as illustrated in equation on (6). Table 2 resumes the enriched results obtained using Symptom ontology (O1 ) and Cardiovascular ontology (O2 ). According to the sketched histograms, we can point out that most of enrichment rates of databases using O1 are higher than O2 . We remark that DB4 has the highest rate for the two ontologies, which explained by the similarity between the components of the ontology and DB4 . It is a surgical database, which is related to the topics of the two used ontologies. Besides DB4 have a small number of tables which proves the enrichment need. In contrast, DB2 have a small rater of changement when we use O2 . The explanation is that DB2 and O2 both are specialized on cardiovascular domain, so it contains most of similar components and does not require a high rate of change. We conclude that the enrichment of databases is based on the relation, between databases and ontologies, and the completeness of databases. 5.2. Qualitative evaluation The qualitative evaluation consists on evaluating the information stored in database. Experts evaluate the resultant database based on five quality criteria 26 : (1) Accessibility: the ability to access to database, (2) Accuracy: the degree to which the data mirrors the characteristics of the real world object or objects it represents, (3) Timely: the time the real world event being recorded occurred, (4) Completeness: the absence of blank and missed values, and (5) Consistency: the absence of difference, when comparing two or more representations of a thing against a definition. We asked twenty experts to judge the 2 3
http://www.obofoundry.org/ontology/cvdo.html http://www.snomed.org/snomed-ct
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
394
Table 2. The enrichment rates for databases using O1 and O2.
Database
E1
DB1 DB2 DB3 DB4 DB5
93 84 145 75 98
E2 103 101 172 104 107
O1
(%) 10.75 20.23 18.36 38.66 9.189
E2 97 86 164 113 109
O2
(%) 4,12 2,32 11,58 33,62 10,09
Fig. 6. Enrichment rates.
information quality of new databases compared to the last version of database. To judge we have two dimensions of satisfaction (high and low). Table 3 presents the number of expert for each dimension and for each criteria. Based on table 3, we conclude that the number of experts, which are satisfied of resulted databases, is higher than the number of experts which ae not satisfied of the quality of databases. Criteria that are more satisfactory for users are accessibity, completeness, and consistency. It proves that users have not problems in the access to databases. Also, it proves that data in databases are correct, complete, and characterized by the absence of ambiguity. Timely criterion is satisfactory for users. However, accuracy criterion is unsatisfactory for some users. This criterion depends to users because accuracy is the closeness of terms used in the database to the terms accepted by users. Furthermore, the selected databases belong to medical domain which characterized by several terms. Accuracy is correlated with context which explains the diversity in the level of satisfaction of users for accuracy criterion. Consequently, terms in database are considered accurate for some users and inaccurate for others. The results of qualitative evaluation show that the enriched databases are satisfying in term of information quality mainly for criteria of accessibility, completeness, and consistency. 5.3. Discussion We can summarize the performance of our approach based on above evaluations of experiments. Both quantitative and qualitative evaluations highlights the efficiency of our automatic approach of database enrichment. Table 2 presents the results of quantitative evaluation which are the rate of enrichment for each database. The obtained results show the performance of our approach for different databases. Then, we evaluate the quality of database after the enrichment. Table 3 presents the level of satisfaction of experts based on five quality criteria. This table shows that
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
Table 3. Level of satisfaction of the experts according to quality criteria for each database.
Information quality criteria
Accessibility Accuracy Timely Completeness Consistency Total
DB1
High 20 10 12 19 15 76
DB2
Low 00 10 08 01 05 24
High 18 12 14 16 14 74
DB3
Low 02 08 06 04 06 26
High 15 11 13 18 17 74
Low 05 09 07 02 03 26
DB4 High 20 09 12 15 17 73
395
DB5 Low 00 11 08 05 03 27
High 18 14 15 17 13 77
Low 02 06 05 03 07 23
experts are satisfying with the enriched database in the most of criteria. Through all the evaluation of experiments, we demonstrate that our approach has enhanced the enrichment of database significantly. 6. Conclusion We proposed an approach to automate the enrichment of database using ontology. The goal was to achieve an automatic way to analyze and compare between the components of ontology and database. We presented in this paper a set of enrichment rules which are divided in two categories: the first category is for the database enrichment using records and the second category is for the enrichment database structure enrichment. We developed a platform to apply rules and we use five medical databases and two medical ontologies. We evaluate the resulted databases with quantitative and qualitative evaluations. The quantitative evaluation shows that the number of database component grows after the application of rules. The qualitative evaluation shows that resulted database meets the quality criteria. As a future work, we will test our approach using voluminous databases and ontologies for different domain. And we will deal with the addition of new semantic relations using the inference rules. References 1. Dobrescu, E.M., PociovliÅteanu, D.M.. Some observations regarding the international databases. Procedia Economics and Finance 2015; 32:841 – 849. 2. Martinez-Cruz, C., Blanco, I.J., Vila, M.A.. Ontologies versus relational databases: Are they so different? a comparison. Artificial Intelligence Review 2011;38(4):271–290. 3. Viale, P., Bora, J.J., Benegui, M., Basualdo, M.. Human endocrine system modeling based on ontologies. Knowledge-Based Systems 2016; 111:113 – 132. 4. Kontopoulos, E., Martinopoulos, G., Lazarou, D., Bassiliades, N.. An ontology-based decision support tool for optimizing domestic solar hot water system selection. Journal of Cleaner Production 2016;112, Part 5:4636 – 4646. 5. Nakhla, Z., Nouira, K.. Ontology based database for computer medical system. In: 2nd Science And Information conference (SAI), IEEE. IEEE; 2014, p. 443 – 446. 6. Damljanovi, D., Agatonovi, M., Cunningham, H., Bontcheva, K.. Improving habitability of natural language interfaces for querying ontologies with feedback and clarification dialogues. Web Semantics: Science, Services and Agents on the World Wide Web 2013;19:1 – 21. 7. Hsu, W., Gonzalez, N.R., Chien, A., Villablanca, J.P., Pajukanta, P., Viuela, F., et al. An integrated, ontology-driven approach to constructing observational databases for research. Journal of Biomedical Informatics 2015;55:132 – 142. 8. Ding, L., Kolari, P., Ding, Z., Avancha, S.. Using Ontologies in the Semantic Web: A Survey. Boston, MA: Springer US; 2007, p. 79–113. 9. Estival, D., Nowak, C., Zschorn, A.. Towards ontology-based natural language processing. In: Proceeedings of the Workshop on NLP and XML (NLPXML-2004): RDF/RDFS and OWL in Language Technology; NLPXML ’04. Stroudsburg, PA, USA: Association for Computational Linguistics; 2004, p. 59–66. 10. Jovic, A., Prcela, M., Gamberger, D.. Ontologies in medical knowledge representation. In: Proceeding of the 29th International Conference on. Information Technology Interfaces, ( ITI’07). IEEE; 2007, p. 535–540. 11. Li, G., Peng, Q.. Using ontology and rough set building e-commerce knowledge management system. In: Proceeding of Advances in Computer Science, Intelligent System and Environment (CSISE 2011); vol. 3. 2011, p. 13–18. 12. Nakhla, Z., Nouira, K.. Ontology maps to database: Ontomapdb approach. In: International Conference on Control, Engineering Information Technology (CEIT’13). IEEE; 2013, p. 103–108. 13. Gali, A., Chen, C.X., Claypool, K.T., Uceda-Sosa, R.. From Ontology to Relational Databases. Berlin, Heidelberg: Springer Berlin Heidelberg; 2004, p. 278–289.
396
Zina Nakhla et al. / Procedia Computer Science 112 (2017) 387–396
14. Chuprina, S., Postanogov, I., Nasraoui, O.. Ontology based data access methods to teach students to transform traditional information systems and simplify decision making process. Procedia Computer Science 2016;80:1801 – 1811. 15. Kamal, H., Fouzia, B.. Semantic enrichment of relational databases. In: Second International Workshop on Advanced Information Systems for Enterprises (IWAISE). 2012, p. 15–19. 16. Campa˜na, J.R., Mart´ın-Bautista, M.J., Medina, J.M., Yager, R.R., Bulskov, H., Christiansen, H., et al. Semantic Enrichment of Database Textual Attributes. Berlin, Heidelberg: Springer Berlin Heidelberg; 2009, p. 488–499. 17. Sujatha, B., Raju, S.V.. Ontology based natural language interface for relational databases. Procedia Computer Science 2016;92:487 – 492. 2nd International Conference on Intelligent Computing, Communication amp; Convergence, {ICCC} 2016, 24-25 January 2016, Bhubaneswar, Odisha, India. 18. LePendu, P., Dou, D.. Using ontology databases for scalable query answering, inconsistency detection, and data integration. J Intell Inf Syst 2011;37(2):217–244. 19. Sun, Z., Finkelstein, A., Ashmore, J.. Using Ontology with Semantic Web Services to Support Modeling in Systems Biology. Berlin, Heidelberg: Springer Berlin Heidelberg; 2007, p. 41–51. 20. Bai, R., Wang, X., Liao, J.. Using an Integrated Ontology Database to Categorize Web Pages. Berlin, Heidelberg: Springer Berlin Heidelberg; 2010, p. 300–309. 21. Wien, J., Knapen, M., Janssen, S., Verweij, P., Athanasiadis, I., Li, H., et al. Using ontology to harmonize knowledge concepts in data and models. 2007. 22. Verykios, V.. Automating the approximate record-matching process. Information Sciences 2000;126(1-4):83–98. 23. Wang, X., Ling, J.. Multiple valued logic approach for matching patient records in multiple databases. Journal of Biomedical Informatics 2012;45(2):224 – 230. 24. Huang, H., Wu, X., Sonachalam, M., Mandape, S.N., Pandey, R., MacDorman, K.F., et al. Paged: a pathway and gene-set enrichment database to enable molecular phenotype discoveries. BMC bioinformatics 2012;13(Suppl 15):S2. 25. An, Y., Khare, R., Song, I.Y., Hu, X.. Automatically Mapping and Integrating Multiple Data Entry Forms into a Database. In: Jeusfeld, M., Delcambre, L., Ling, T.W., editors. Conceptual Modeling ER 2011; vol. 6998 of Lecture Notes in Computer Science. Springer Berlin Heidelberg; 2011, p. 261–274. 26. Fehrenbacher, D.D.. Perceptions of information quality dimensions from the perspective of commodity theory. Behavior and Information Technology 2016;35(4):254–267.