SylvaDB: A Polyglot and Multi-backend Graph Database Management System

SylvaDB: A Polyglot and Multi-backend Graph Database Management System Javier de la Rosa1, Juan Luis Su´ arez1and Fernando Sancho Caparrini2 1CulturePlex Lab, Dept. Modern Languages and Literature, Western University, Western Rd., London, Canada 2Ciencias de la Computacin e Inteligencia Artificial, CulturePlex Lab, University of Seville, Seville, Spain Keywords: Graph Databases, Polyglot Backends, Network Research. Abstract: This paper presents SylvaDB, a graph database management system designed to be used by people with no technical knowledge. SylvaDB is based on flexible schema definitions and has been developed taking into account the need to deal with semantic information. It relies on the mathematical notion of property graph. SylvaDB is an open source project and aims at lowering the barrier of adoption for anyone using graph databases. At the same time, it is robust and scalable enough to support collaborative large projects related to knowledge management, document archiving, and research. 1 INTRODUCTION After several years of work in an international and multidisciplinary project on cultural transfers, whose objectives included the analysis of several datasets of various types of cultural objects, we came to the conclusion that both the classic relational databases and the total dependence of researchers on programmers to deal with their data was detrimental to the purposes and objectives of the project. We came to this conclusion for several reasons. First, we realized that non technical researchers presented a strong tendency towards resigning their own research autonomy and handing in all the decisions on database design and implementation to the professional programmer. However, this attitude was usually accompanied by the unsatisfactory results of the collaboration, as humanists, for example, were not able to extract or analyze their own results. And second, as the projects evolved, non technical researchers would come back to the programmers demanding changes in the structure of their databases, asking for new features during incremental design, corrections of mistakes in the database design, or modifications to reflect changes in the structure of the real world artefacts modelled in the database. But despite of the efforts of the database community to solve schema modification and evolution, as noticed by (Roddick, 1992), there is still a gap between the database administrator who manages to solve this issues, and the end user who needs them solved. In an attempt to improve the interaction with databases, research efforts have been dedicated to provide graphical interfaces to relational databases such as HIBROWSE (Ellis et al., 1994), Query-ByExample (Zloof, 1975), or Santucci and Palmisano’s Visualiser (Santucci and Palmisano, 1994). Also, some other works reveals moves from these complex graphical interfaces to more expressive and closer to natural languages alternatives (Hendrix et al., 1978; Wolfe et al., 1992; Popescu et al., 2003). However, there is no evidence of providing better and easier administration tools for non technical users that allow them to manage their data. On the other hand, one of the most active areas of research at this moment is the study of networks (Knoke et al., 2008), what has brought the need to manage information with inherent graph-like nature (Angles and Gutierrez, 2008). Therefore, after surveying the landscape (Vicknair et al., 2010), we decided to build our own database management system, SylvaDB1, in order to implement the graph model instead of the relational and to provide researchers with the tools they need as well as to reduce the dependency on programmers and database administrators. 2 DESIGN PRINCIPLES The main goals that guided the design of this graph 1http://sylvadb.com database management system were: •to provide technological autonomy to non technical researchers, •to allow for modifications and evolutions of the schema at any time of the life cycle of the database, •to be scalable and therefore useful for both the single researcher working on a relatively small project and the requirements of large teams, •to manage different datasets under the same system, •to allow collaborative work on single databases, •to support as many as possible types of objects, •to present the information by using an intuitive interface. Out of our own experience in Complex Systems and Cultural Analysis, we realized that topics and relations among them are at the same level of interest. Consequently, and from previous goals, the following conceptual and technological features were mandatory for us: •No tables, only objects and relations. •Arbitrary number of attributes on objects and relationships. •Support for multimedia content natively (image, audio, video and documents). •Users and permissions management. •Ability to perform complex queries to retrieve and analyze information. •Interactive graphic visualization of the content. •Connection to external tools for visualization and analysis. Django uwsgi memcached Nginx Static files (Amazon S3) SylvaDB (Amazon EC2) PostgreSQL (Heroku) Graph Instance 1 (Amazon EC2) Graph Instance N (Amazon EC2) Payments (Stripe) Neo4j HA (Amazon EC2) Neo4j Node Neo4j Node Neo4j Node Neo4j Node Figure 1: General Architecture of SylvaDB. to guarantee compatibility with as many of them as possible, we store the schema (if any) in a relational database while all data is actually stored in the graph database backend; properties for both nodes and relationships are also stored in the graph backend; and the set of media files and other static files are stored in Amazon Simple Storage Service (S3)2. SylvaDB’s architecture, as depicted in Figure 1, is divided into several parts in terms of machine isolation, so any of them can be replaced or clustered with no major problems3. In short, there is a main HTTP server that serves the Django application SylvaDB through an uwsgicompliant Python process. All statics files (including user uploads and client-side Javascript and styles) are handled in Amazon S3 instances. The relational database to manage users, sessions, permissions and graphs metadata like their schemas, is managed in PostgreSQL instances by the Platform as a Service provider Heroku (Malliga, 2012). Actual data for graphs in the shared graph backend is stored in a Neo4j High Availability (HA) cluster, and graphs in custom instances may be in different kinds of graph backends as described in section 3.2. On the other hand, both low level and view level cache are implemented in a distributed memcached environment (Fitzpatrick, 2004). This setup confers reliability and performance to SylvaDB at the time that makes scalability a problem easy to solve. 3.1 HTTP Server Nginx4is an asynchronous HTTP and reverse proxy server that natively includes support for upstream 2https://aws.amazon.com/en/s3/ 3SylvaDB also gives support to services such as registration and messaging, but we have omitted them in the diagram, as they play an auxiliary role in our architecture. 4http://nginx.com/ 3 ARCHITECTURE SylvaDB relies on the paradigm of Database as a Service (DBaaS), as described by Hacigumus (Hacigumus et al., 2002). Written using the Python programming language, and built through the ModelView-Controller web framework Django (Holovaty and Kaplan-Moss, 2009), SylvaDB is a web platform that can be used in modern browsers such as Google Chrome or Mozilla Firefox. In SylvaDB a graph database consists of a schema with information on data types and properties for nodes and relationships, and the real data stored on them. Because some graph backends are schema-free and some others are schema-restricted, and in order servers speaking the WSGI5protocol, such as uWSGI6does for Django. The main advantage of the asynchronous approach is scalability. In a processbased server, each simultaneous connection requires a thread which incurs significant overhead. An asynchronous server, on the other hand, is event-driven and handles requests in a single, or very few threads. While a process-based server can often perform on par with an asynchronous server under light loads, if heavier loads occur the performance degrades significantly as RAM consumption increases. 3.2 Graph Model At the core of the system, we use the ObjectRelational Mapping (ORM) (Burza´ nska et al., 2010) provided by Django to manipulate the programming objects in the code, and a set of mixins to abstract graph backends (Esterbrook, 2001). Figure 2 illustrates a simplified UML diagram of the elements implied in the Graph model. Django terminology uses model when refering to Python classes that inherit from the base class that does the mapping between the code and the database; it provides some helpers to avoid the manual manipulation of tables, foreign keys and many to many relationships. To emulate a similar behaviour in relation to the graph backends, and to make it easier to develop SylvaDB, we wrote some mixins to abstract all the graph databases primitives. In the diagram we omit inheritance from Django models. Therefore, SylvaDB makes use of Graph object instances that cover both the schema and the backend. Our graph structure is based on the property graph, i.e., a multigraph data structure where graph elements, nodes and relationships, can have properties (attributes) and can be typed. Formally, the property graph is defined as a tuple G= (V,E,P,DV,DE,νV,νE), where Vis a set of vertices, Eis a multiset of directed edges, Pis a domain of properties, DV,DEare the domain of allowed property values for vertices ad edges, respectively, νX: X×P→Pf(DX)(X=Vor X=E) is a function that maps properties of elements of Xto their values (Pf(DX)is the collection of finite subsets of DX, meaning that a property can be associated with multiple items from DX). 3.2.1 Flexible Schemas The first component of a Graph object is the schema, 5http://wsgi.readthedocs.org/ 6http://uwsgi-docs.readthedocs.org/en/latest/ Protocol.html <<object>> Graph Backend <<django.db.models.Model>> Relational Backend Graph +data: OneToOneField +schema: OneToOneField <<mix-in>> GraphMixin +nodes +relationships DataSchema <<facade>> NodeManager <<facade>> RelationshipManager Instance BaseType RelationshipType Node Relationship 1 1 1 1 NodeType 1 * 1 * GraphDatabase BlueprintsGraphDatabase Neo4jGraphDatabase 1 1 Property 1 * 1 * Figure 2: Simplified UML Diagram. albeit it is not required of the graph to have one. Together with the fact that some users may need no schema in their data structure, the idea behind this flexible schema is to design the platform as independent as possible from the graph backend. Since some graph bakends may require a schema and others do not, we move the schema out of the graph backend and made it optional. Even when this decision may involve a manual managing of all triggers commonly handled by relational databases, at the same time provide us with a powerful and fine grained control over schema evolution (the ability of a database system to respond to changes in the real world by allowing the schema to evolve (Roddick, 1992)). Relational databases usually have fixed schemas, but SylvaDB does not have this limitation because is based on graphs, and schema information is only tied to graph data through the application logic. SylvaDB makes users owners and responsible for their schema evolution, at the time it provides them with the tools to decide. Before adding nodes to a graph, a schmea must be defined by creating node types and relationship types. By using web forms generated by SylvaDB, the user is able to add properties to the types and build the schema. Every time a user makes a change in a graph schema, he is prompted to choose what to do depending on the operation he wants to perform: •If the user renames a node type or relationship type, nothing will happen since types have internal numerical identifiers. •If the user renames a property, SylvaDB can keep the previous property name internally in the item (node or relationship), can remove the old property from the item and consider the new property as new and empty, or can try to rename the last property with the new one. •If the user removes a node type or relationship type, will be asked to decide if all related items must be removed on cascade or just the type. •If the user removes a property, SylvaDB can also keep the value internally in the item or remove it from all the items affected. 3.2.2 Polyglot Graph Backend The second component of a Graph object is a (mandatory) graph backend. From its inception, SylvaDB was designed to support different technologies as backend, what we call polyglot graph backend, since there exist different graph databases providers for different needs. For example, Titan7is intended for distributed and massive-scale graphs, whereas Neo4j is faster but does support a slightly lower number of nodes. Currently, the graph databases providers landTable 1: Summary of graph databases, query methods and Python bindings. Graph Query Python Database Method Binding Cypher, Blueprints, Neo4j Gremlin, Native, Traversal REST OrientDB Cypher Blueprints Traversal HyperGraphDB HGQuery No Traversal DEX Traversal Blueprints Titan Gremlin Blueprints InfiniteGraph Gremlin No scape is not as huge and diverse as the relational, and unfortunately most of them are only suitable to be used in Java language programming. To overcome this limitation, and having in mind 7http://thinkaurelius.github.io/titan/ that SylvaDB base code is Python, we resorted to REST (Representational State Transfer) interfaces (Fielding, 2000). Specifically, SylvaDB makes use of the Blueprints API8, a generic graph Java API that binds to various graph database backends. Many graph processing and programming frameworks are built on top of it, e.g., Gremlin9, a graph traversal language, and Rexster10, a graph server that exposes any Blueprints graph through a REST interface. Not all graph database backends have a built-in support for the Blueprints API, as the Table 1 shows, and others have their own interfaces, like Neo4j. Figure 3 shows how, on top of these REST interfaces (Blueprints and Neo4j), SylvaDB defines a set of abstraction layers to operate with them exposing a common and unique set of methods and attributes, the GraphDatabase interface. Any class able to implement this interface, will be suitable as a graph backend in SylvaDB. Nevertheless, only Neo4j and Blueprints-compliant graph databases are available now, thanks to client libraries like neo4j-rest-client11 and pyblueprints12, respectively. Neo4jOrientDB SailRDFDEXTitan Blueprints API REST neo4j-rest-clientpyblueprints GraphDatabase Figure 3: Polyglot graph backend diagram. Since each graph backend run on its own service, SylvaDB is able to deploy backends on demand, by using Amazon CloudFormation13 templates, and offer them to users with different needs. 3.3 Queries For the purposes SylvaDB has been built, easy ways 8http://blueprints.tinkerpop.com 9http://gremlin.tinkerpop.com 10http://rexster.tinkerpop.com 11http://pypi.python.org/pypi/neo4jrestclient/ 12http://pypi.python.org/pypi/pyblueprints/ 13http://aws.amazon.com/cloudformation for retrieval information from data was as important as data itself, making it mandatory to provide tools to build powerful queries to the user. However, although it is not a new topic, recent research and implementations are finally reaching a tiping point with regards to providing natural language interfaces for querying databases (Popescu et al., 2003). In this context, SylvaDB tries to be a step ahead by generating schema-based grammars to process users’ queries. Although the grammar depends on the graph schema and is domain-specific and quite limited yet, it is able to process approximate natural language queries as inputs and produce backend-specific queries as outputs. As an example, let’s suppose we have a graph called ”Workers” with a Neo4j backend, which supports Cypher Query Language14, then the query: Person who lives in country with name Spain will produce the following Cypher query: START node=node:graph(label=’Workers’) MATCH person-[:lives_in]->country WHERE country.name = ’Spain’ AND person.type = ’person’ AND country.type = ’country’ RETURN person Unfortunately, for more complex queries (such as returning nodes in the middle of a path) or to support schema-free graphs, the only way around is by using the query language available for the specific backend. After having explored the options to implement a visual query system (Catarci et al., 1997), it is clear that SylvaDB has considerable room to improve in this area. 3.4 Visualization For graphs with defined schemas, paginated tablebased views are provided for each type in the schema. Since the data model is based on graphs, SylvaDB may provide a proper way to visualize users’ data in a more graphical manner. There are so far two different kinds of visualization: the first is our own development, and it allows users to expand relationships by clicking on the nodes. However, for graphs with more than a couple of thousand nodes this visualization, although more complete, is not very useful. The second one is based on sigma.js15, an open-source lightweight JavaScript library to draw graphs, and using the HTML5 canvas element16. 14http://docs.neo4j.org/chunked/stable/ cypher-query-lang.html 15http://sigmajs.org/ 16http://w3.org/TR/html5/embedded-content-0.html For other general purposes, the system provides exporting functionality to GEXF17 files, a graph format file defined for Gephi (Bastian et al., 2009). 3.5 Collaboration Fine grained permissions is also an important feature that SylvaDB includes. Implemented to graph object level, graph owners can grant up to 9 different permissions for their graphs to other users: •over the graph: view, change collaborators, change graph properties (information description and visibility), •over the data: all CRUD operations for nodes and relationships, •and over the schema: view and change. 4 APPLICATIONS Thanks to the versatility of the conceptual storage data type supported, the property graph, the uses of SylvaDB vary from a tool for standard storage, or an implementation system for ontologies, glossaries, conceptual maps and relaxed topic maps, to an analysis tool useful to find non explicit relations between topics or discover hidden knowledge through queries that traverse the graph. Below follow three well studied cases that have proven how useful SylvaDB can become. 4.1 Formal Ontologies: Baroque Paintings Network This research on baroque paintings is part of a 7-year project that deals with the issues of cultural transmission and assimilation and community formation during the baroque period in the territories of the former Hispanic Monarchy. One of the important aspects of this historical case lies in the fact that the political structure that supported or accompanied the studied cultural transfers reached almost global proportions, from Europe to American and Asia. Thus, it offered an excellent benchmark to test some of the prevalent theories in the field taking advantage of a dataset that included over 13,000 paintings, 1636 creators, 405 series of paintings, 195 schools and 2482 geographical locations from different territories and cultural areas. So far, extensive research have been done on three main issues. First, an analysis of the data set to answer questions related to the formation and sustainability of large cultures, the semantic content of the 17http://gexf.net/format/ San Andrés San Mateo El Niño Jesús y San Juan Santa Agueda Portrait of a Lady (Retrato de una dama) Gaspar Sánchez Juan Bautista de Urquiola y Elorriaga Ildefonso Muños Tirado Bodegón del cardo Bodegón con dulces y recipientes de cristal Bodegón de frutas y hortalizas Bodegón Nuestra Señora de Copacabana Nuesta Señora del Pilar con un monje franciscano y uno dominico La virgen de Almudena con Carlos II y la reina de España Felipe III Retrato ecuestre del duque de Lerma Aguador de Sevilla Vieja friendo huevos Tres músicos Don Luis de Góngora y Argote Caballero Anciano Isabel de Valois, tercera esposa de Felipe II Caballero joven Jerónimo de Cevallos La infanta Isabel Clara Eugenia El archiduque Alberto de Austria La reina Ana de Austria, cuarta esposa de Felipe II Retrato de caballero Cabeza de anciano Francisco Pacheco o Retrato de hombre con gola Ana de Austria, esposa de Luis XIII La familia del pintor María de Medici, reina de Francia Retrato de hombre (Autorretrato de Velazquez) La mulata Tres hombres a la mesa Dos jóvenes a la mesa Felipe IV Retrato de caballero, Juan de Fonseca Felipe IV El conde duque con la cruz de Calatrava Fiesta de bodas campesina El patizambo Giovan Carlo Doria a caballo Retrato doble Alessandro Giustiniani con traje de senador y su esposa, dama de la casa Spinola Los síndicos del gremio de los pañeros Sir George Villiers, futuro duque de Buckingham, y su esposa Katherine Manners retratados como Venus y Adonis Retrato de familia El Sacamuelas La familia Lomellini Autorretrato (Bernini) Don Juan de Austria Felipe IV Retrato del Conde de Villalonga Lope de Vega Retrato de progenitor El Primo Retrato de damaRetrato masculino Familia de Mendigos Retrato de hombre Niño desnudo Retrato del príncipe Felipe Manuel de Saboya Retrato de doña María de Médicis Felipe III Retrato de Lope de Vega Gabriel Téllez (seudónimo de Tirso de Molina) Agustín Moreto y Cavana Retrato de Felipe de Médicis, hijo de Francisco I Cosme II Cristina de Lorena, duquesa de Florencia Self-Portrait (Autorretrato de Luis Tristán) The Sculptor Gregorio Fernandez (El escultor Gregorio Fernández) Autorretrato (Vicente Carducho) Retrato de anciano Retrato de George Gage con dos Asistentes Autorretrato con su esposa Isabel Brant Retrato de Elena Grimaldi Cattaneo Self-Portrait (Autorretrato de Pedro Orrente) Felipe III como Gran Maestre de la Orden del Toisón Retrato de Jacobo I de Inglaterra Retrato de don Francisco Gómez de Sandoval y Rojas, duque de Lerma Retrato de Antonio de Servás Queen Margaret and Princess Ana as the Annunciation (La reina Margarita y la Princesa Ana en la anuciación) James Hamilton, conde de Arran y primer duque de Hamilton Queen Margaret of Austria (La reina Margarita de Austria) Portrait of Philip IV and Ana (Retrato de Felipe IV y Ana) Portrait of Alfonso -el Caroand Ana Margarita (Retrato de Alfonso -el Caroy Ana Margarita) King Philip III of Spain (Rey Felipe III de España) Francisco Gómez de Sandoval y Rojas, Duke of Lerma (Duque de Lerma) Portrait of Lorenzo van der Hamen y León (Retrato de Lorenzo van der Hamen y León) Portrait of a Monk (Retrato de un monje) Autorretrato Retrato de Maria de Medicis Retrato de Virginia Centuriona Fray Hernando de Rojas El conde duque de olivares La Reina Margarita de Austria El Infante don Alfonso el caro Retrato de joven dama Felipe III La Duquesa de Módena Cabeza de muchacha Retrato de Agustín Moreto Copia del retrato de Doña Mariana de Austria de Diego de Velázquez del Museo del Prado, nº 1191 Idolatría de SalomónDos genios entre guirnaldas y ramos de flores, Amorcillos con guirnaldas Santa Margarita de Escocia Retrato de Jean de Montfort Don Adrián Pulido Pareja Copia de las Meninas Retrato de Caballero (Fransisco Pacheco?) Don Pedro Calderón de la Barca Copia de Retrato de Doña Mariana de Austria de Juan Carreño de Miranda del Museo de Viena Sacrificio de Isaac Lope de Vega Retrato de clérigo Retrato de hombre Retrato masculino Retrato de dama David con la cabeza de Goliat Brigidad Spinola Doria Isabelle Brandt Maria Serra Pallavicino Autorretrato Retrato de Isabelle Brandt Retrato del cardenal Guido Bentivoglio Retrato de la marquesa Balbi Retrato de Isabel la Católica Retrato de don Cristóbal Suárez de Ribera View and Plan of Toledo Retrato de niño Arcángel san Miguel con donante indígena Virgen del Rosario con donante indígena Retrato de Jorge Manuel Theotocopuoulos Retrato de Simón Ruiz Retrato de doña Mariana de Paz Pineda David con la cabeza de Goliat Virgen de Guápulo Virgen del Rosario con Santo Domingo y San Francisco La adoración de los Reyes Magos La Resurrección La anunciación La Crucifixión Pentecostés Cristo abrazado a la Cruz Fraile trinitario Adoración de los Reyes Adoración de los pastores Adoración de los pastores El Salvador Santiago Susana y los viejos Ofrenda a Ceres La serpiente de metal La venerable madre Jerónima de la Fuente ( o Sor Jerónima de la Fuente) El Prendimiento de Cristo La Coronación de espinas La Verónica La visita de Baco al poeta Icario Retrato de San Francisco Magdalena Penitente Cristo crucificado contemplado por la Virgen María y San Francisco Santa Catalina de Siena La Virgen y el Rosario con Santo Domingo y Catalina de Siena Nuestra Señora del Rosario Nuestra Señora del Pilar de Zaragoza Nuestra Señora del Sagrario de Toledo Nuestra Señora de La Salceda La venerable madre Jerónima de la Fuente ( o Sor Jerónima de la Fuente) Cristo atendido por los ángeles Cristo en casa de Marta y María Don Cristóbal Suárez de Ribera La Virgen imponiendo la casulla a san Ildefonso Cena de Emaús Cristo flagelado Cristo flagelado La coronación de la Virgen Crucifixión de San Pedro Salomé El tocador de Venus Martyrdom of Saint Hippolytus (Martirio de San Hipólito) Huida a Egipto Paisaje con Mercurio y Argos La Virgen de Guadalupe Et in Arcadia ego Muerte de la Virgen Cena en Emaús Resurrección de Lázaro Vocación de san Mateo La visión de san Francisco Martirio de santa Rufina y santa Segunda La Virgen presenta el Niño a santa Francisca Romana San Pedro liberado de la cárcel Sagrada Familia Incredulidad de santo Tomás La Virgen despierta al Niño Pentecostés Desposorio místico de santa Catalina Virgen del Rosario Siete obras de misericordia Castigo de Cupido David con la cabeza de Goliat Sansón y los filisteos Martirio de san Pedro Rapto de Proserpina San Juan Bautista Las mujeres en el sepulcro Marte y Venus Ecce Homo Bautizo de los cuatro señores de Tlaxcala Cristo de la columna (Jesús de la Humildad y la Paciencia) El Niño Jesús con los símbolos de la pasión Ángel Zadquiel Ángel Dominio Ángel Baraquiel Arcángel San Miguel Letiel Dei Gabriel Dei Ángel de la Guarda Nuestra Señora de Pomata Concierto angélico Prendimiento de Cristo Adoración de los Reyes Magos Moses (Moisés) The Adoration of the Magi (La Adoración de los Magos) St. Mary of Egypt (Santa María de Egipto) The Archangel Barachiel, with Harquebus (El Arcangel Baraquiel con rifle) Arcángel Uriel, Fuego de Dios Arcángel Seactiel, Oración de Dios Arcángel Miguel, Quién como Dios Arcángel Baraquiel, Bendición de Dios Sagrada Familia con Santa Isabel y San Juanito David y Betsabé Santiago del Mayor Martirio de San Sebastián Oración del huerto San Ignacio curando a los apestados Virgen de la Soledad San Mateo Cristo Yacente Seactiel (Oración de Dios) Angel Custodio Jesús de la Humildad y la Paciencia La adoración de los pastores La Inmaculada La Virgen de la Merced entregando el hábito de la orden a don Manuel Alonso, Duque de Medina Sidonia La Santa Cena Calvario Sagrada Familia Nativity (Natividad) Transfixión de Santa Teresa Santo Tomás de Aquino La madre Jerónima de la Fuente San Miguel de los Santos Arrepentimiento de San Pedro La Verónica Cristo (puerta de sagrario) Concepción y coronación de la Virgen Inmaculada Concepción Virgen de Guadalupe Virgen con el Niño y San Juanito El taller de Nazareth San Isidro Labrador Santo Tomás San Juan Evangelista Julián Romero y su santo Patrono San Francisco sostenido por un ángel Lucha de San Jorge y el dragón San Sebastián Piernas de San Sebastián San Francisco confortado por un ángel músico Santa Margarita resucita a un joven Desposorios místicos de Santa Catalina El Apóstol Santiago El Mayor San Pedro liberado por un ángel La curación de Tobías La Inmaculada Concepción San Juan Evangelista San Pablo San Pedro y San Pablo Virgen Inmaculada con San Francisco y San Antonio Santa Ursula y las once mil Vírgenes Los milagros de san Ignacio de Loyola San Pedro liberado de la cárcel Sagrada Familia con santa Isabel, san Juanito y un ángel Santo Tomás de Aquino fulminando al demonio con su pluma Virgen de Chiquinquirá con el Niño con San Antonio y San Andrés St. Joseph and the Infant Jesus (San José y el Niño Jesús) Virgen de Chiquinquirá con el Niño, San Antonio y San Andrés St. Francis of Assisi (San Francisco de Asís) St. Nicholas of Bari (San Nicolás de Bari) San Bartolomé San Jose y el Niño San Ignacio en la carcel en Alcala de Henares Las Lagrimas de San Pedro San Buenaventura Jesus crucificado con la Virgen Maria y San Juan Cabeza de Pablo La Visión de San Ignacio en el camino de Roma San Ignacio recibe en la compañia a San Francisco Javier San Ignacio despide a San Francisco de Javier La muerte de San Ignacio San Pedro San Mateo San Pablo San Judas Tadeo San Felipe Santo Tomas San Andrés Santa Casilda Cristo de Los Temblores The Decent from the Cross (el descendimiento de la cruz) Los Desposorios de la Virgen Adoración de los pastores San Pedro Nolasco asistido por dos ángeles San Bartolomé San Fausto Labrador San Juan Evangelista San Ignacio de Loyola Busto de Santa El abrazo de Santo Domingo y San Francisco San Agustín San Andrés San Francisco de Borja San Pedro San Pedro Nolasco Santo en oración. San Pedro Un Santo. San Zacarías Verdugo con la cabeza de san Juan Bautista Santa Catalina San Jerónimo Santa en éxtasis El martirio de Santa Catalina de Alejandría Los desposorios místicos de Santa Catalina San Francisco y el lego El apóstol San Mateo San Francisco de Asís San Pablo San Pedro San Sebastián curando por las Santas Mujeres St. Francis and Brother Leo Meditating on Death (San Francisco y hermano Leo meditando sobre la muerte) The Penitent St. Jerome in his study (El penitente, San Jérome en su cuarto de estudio) San Francisco en oración San Diego de Alcalá Vestición de san Guillermo de Aquitania Predicación de San Juan Bautista San Agustín entre Cristo y la Virgen La Sagrada Familia con Santa Ana y San Juanito San Pedro Nolasco despidiéndose de Jaime I el Conquistador Crucifixión de San Pedro Sagrada Familia con Santa Ana San Joaquín y Santa Ana contemplando a la Virgen con el Niño El tránsito de San Isodoro El martirio de San Andrés Santa Ana enseñando a leer a la Virgen Martirio de San Ramón Nonato Resurrección de Santa Leocadia Martirio de Santa Úrsula y las once mil vírgenes La apoteosis de San Hermenegildo El éxtasis de San Francisco Javier San Pedro liberado por el ángel (La liberación de San Pedro) San Jerónimo penitente Crucifixión de San Pedro Crucifixión de San Pedro San Cristóbal Los desposorios místicos de Santa Catalina Santa Catalina de Alejandría en prisión San Pedro penitente San Pedro penitente Las Lágrimas de San Pedro San Pedro y San Fransisco San Sebastián Martirio de San Bartolomé San Francisco de Asís penitente San Pedro penitente San Juan Bautista en el desierto San Francisco de Borja San Juan Evangelista en la isla de Patmos San Julián de Cuenca San Agustín San Felipe Neri San Jerónimo Cabeza del Bautista Desposorios de Santa Catalina Saint Francis Embracing the Crucified (San Francisco abrazando a Cristo Crucificado) The stigmatization of Saint Francis (la estigmatización de San Francisco) Santa Humbelina the vision of Saint John (la visión de San José) Estigmatización de San Francisco de Asís Saint Jerome as Scholar (San Jeronimo como cardenal) Saint James (Santiago el mayor) (San Jaime) The stigmatization of Saint Francis (la estigmatización de San Francisco) The Visión of Saint Francis of Assisi (La Visión de San Francisco de Asís) The Ecstasy of Saint Francis of Assisi: The Vision of the Musical Angels (éxtasis de San Francisco de Asís: La Visión de los ángeles músicos) Martirio de San Lorenzo San Lorenzo Santa Catarina salvada del Martirio por un Angel Santo Tomás de Aquino San Juan de Dios San Pedro Martirio de Santiago el Menor San José con el Niño Jesús San Gabriel Arcángel San José La estigmatización de San Francisco La asunción de la Virgen Las tentaciones de San Antonio San Juan Bautista San Pedro Cristo con San Pedro, Santiago y San Juan San Ambrosio San Pedro La Sagrada Familia La Sagrada Familia y santa Catalina, San Carlos Borromeo ante Cristo muerto Descanso en la Huida a Egipto Copia de San Jerónimo de Ribera del Museo Nacional de Arte Antiga de Lisboa Santa Teresa. Autorretrato de pintor / San Cosme Copia de San Esteban de José de Ribera del Museo Nacional de Bellas Artes de La Valletta San Pedro liberado por un ángel San Jerónimo y un rabino Paisaje con San Juan Bautista niño dormido Aparición de la Virgen a San Jacinto de Polonia Cabeza de San Juan Bautista Santa Edith Santa Helena Santa Walburga Sagrada Parentela (Alegoría de la Inmaculada) La Virgen con el Niño y Santos, y el Canónigo Bartolomé Cairasco Santiago en la batalla de Clavijo Visión de San Bernardo o Lactación de San Bernardo El Martirio de San Serapio San Ignacio de Loyola ante la Inmaculada y la Trinidad Tránsito de San Isidoro Anunciación de la Virgen y Encarnación del Verbo Presentación de la Virgen en el Templo Jesús camino del Calvario Santa Ana enseñando a leer a la Virgen San José con el Niño La Dolorosa San Juan Evangelista San Juan Evangelista en la Isla de Patmos San Pedro Nolasco Aparición de la Virgen a san Simón de Rojas Santo Domingo de Guzmán San Agustín San Juan Bautista San Ambrosio San Pedro Nolasco El Salvador Copia o imitación de San Francisco meditando de rodillas de El Greco San Ambrosio Copia de San Francisco en extasis de El Greco. Noli me Tangere Estigmatización de San Francisco Virgen con el Niñ0 y san Juanito Santos Mártires en gloria San Francisco de Asís en éxtasis La Virgen de los Carmelitas de Malinas San José con el Niño San Bartolomé Santa Bárbara El Niño Jesús apareciéndose a San Antonio de Padua San Pedro San Mateo y el ángel La vocación de san Mateo El martiro de san Mateo El martirio de san Pedro La conversión de San Pablo Santa Cecilia y el ángel La Vocación de san Mateo San Juan de la Cruz Santa Teresa de Jesús Santa Teresa de Jesús Paisaje con san Jerónimo Apostle St Peter Apostle St Andrew Apostle St John the Evangelist Apostle St Philip Apostle St Bartholomew Apostle St Matthew Apostle St Thomas Apostle St James the Less Apostle St Thaddeus (Jude) Apostle St Simon Apostle St Paul Cristo como Salvador (Christ as Saviour) San Francisco en éxtasis San José con el Niño Santa Maria Egipciaca Tránsito de San Hermenegildo San Agustín San Simón San Francisco y el hermano León sobre la muerte San Francisco Santo Domingo San Lucas y San Marcos San Juan y San Mateo Conversión de San Pablo San Antonio Abad El martirio de San Ramón Nonato El Martirio de San Lorenzo El martirio de Santa Catalina El Martirio de Santa Ursula y las once mil vírgenes El triunfo de San Gregorio Santo Tomás de Villanueva San Pedro Nolasco embarca para redimir cautivos Aparación de la Virgen a San Ramón Nonnato Sagrada Familia con San Joaquín y Santa Ana Sagrada Familia San Juan de la Cruz, destinado a las Indias Santo Tomás San Felipe San Pablo Vocación de San Pedro y San Andrés Los santos Juanes Sueño de la madre de San Eloy San Francisco en oración. Tránsito de san José San Benito Santa Escolástica San Benito Abad San Pedro Santa Isabel Reina de Portugal San Matías San Bartolomé San Pedro San Simón Los desposorios místicos de Santa Catalina Predicación de San Andrés San Juan Evangelista San Juan Batista San Roque San Sebastián San Fransisco de Asis La santa cena Retablo del Calvario San Luis Rey de Francia San Jerónimo San Francisco de Asis Bodegón de caza y frutas Bodegón Florero Naturaleza muerta Cisne sobre alfombra Bodegón Florero Bodegón con objetos de orfebrería Bodegón de caza, hortalizas y frutas Bodegón de hortalizas Bodegón de frutas Bodegón Still life with Fruit and Vegetables (Bodegón de frutas y hortalizas) Bodegón de frutas Bodegón Bodegón con cisne Bodegón con cajitas de dulces Naturaleza muerta con volatería Cestas de ciruelas e higos y un melón sobre una repisa con pimientos, uvas y membrillos colgados Naturaleza Muerta Florero Bodegón de hortalizas y frutas Bodegón de cocina Florero y frutero Plate of pears (Plato de peras) Still life with Game Fowl (Bodegón de juego de aves) Bodegón Bodegón de frutas y caza Bodegón de manzanas Bodegón Bodegón Bodegón de cocina con frutas y vegetales y mujer Gallo Bodegón Still life with Quince, Cabbage, Melon, and Cucumber (Bodegón de Membrillo, repollo, Melón, pepino) Still life with sweets (Bodegón con dulces) Still life (Plate with Bacon, Bread and Wine (Bodegón con tocino, pan y vino) Still life with Sweets (Bodegón con dulces) Bodegón de frutas Una Perdiz Frutero Frutero y cestillos Frutero y dulces Bodegón Jacob Conjuring Laban's Sheep (Jacobo apareciendo con las ovejas de Laban) Retrato del Barón de Dietrichstein Luis de Velasco Retrato del archiduque Rodolfo de Austria Hernán Cortés Diego Vargas Calderón Retrato de la Condesa de Goldstein Pedro Sebastián de Bolívar y Mena Princesa Margarita Gonzaga García de Carbajal Juan González Juan Bautista de Urquiola y Elorriaga Retrato de la baronesa de Dietrichstein Andrés de Carbajal y Tapia Un matemático Pedro de Barrientos Lomelín Bartolomé Pérez de la Canal Diego Carrillo Mendoza Pimentel Rodrigo Pacheco y Osorio Alonso de la Veracruz Juan Bohorques Ramírez Retrato de Infanta Diego Fernández de Córdoba Dama de la época de Felipe III Gerónimo de Cárcamo Hernán Cortés Juan Cano Bartolomé González Soltero, fray Retrato del archiduque Ernesto de Austria Retrato ecuestre de Pedro Antonio Fernández de Castro (X Conde de Lemos, Virrey del Perú) Juan de Aguirre Juan de Mendoza y Luna Gregorio López Limpiadora de Cobre Pedro Cristóbal Plancarte Retrato de Dama Francisco López de Solís Miguel de Poblete Juan de Cervantes Diego de Contreras Bartolomé de Ledesma Gregorio López y Francisco de Losa Bernardino de Sahagún Nicolás de la Torre Francisco de Armentia Gil y Paulucio de la Trinidad Juan González Juan Millá de Poblete Luis de Sifuentes Juan Díez de Arce Pedro Gutiérrez de Pisa Ildefonso Muñoz Francisco de Siles Isidro Sariñana y Cuenca Manuel Fernández de Santa Cruz y Sahagún Francisco Rodríguez Santos Alonso Enríquez de Toledo y Armendaris Francisco de Contreras Ramírez y Zepeda García Guerra, fray Pedro de Gante Juan González García Guerra Juan de Palafox y Mendoza Juan Díaz Arce La cabeza del Virgen María (Head of the Virgin Mary) Santa Clara con Sor ? o Santa? Juana de la Cruz y Madre Luisa Carriona Francisco Naranjo Rodrigo Maldonado Nicolás del Puerto y Salgado Santa Clara San Jerónimo Santo Tomás de Villanueva (copia de Murillo) Martirio de San Bartolomé San Jerónimo Magdalena Penitence San Diego de Álcala San Juan Evangelista San Joaquín Santo Tomás de Aquino Vicente de San José, fray Inmaculada Carmelita con San Francisco y Santa Clara San Francisco recibiendo los Estigmas Virgen con el Niño entre San Francisco y Santa Clara San Eliseo San José San Pedro San Juan Evangelista y la visión de la mujer del Apocalipsis Anunciación a san José Aparición de la Virgen a san Ignacio de Loyola Concebida Liberación de san Pedro Meditaciones de san ignacio sobre el infierno San Ignacio de Loyola Santiago el Mayor Santo Tomás Apóstol Virgen con el Niño y San Bruno Santa Clara Fundadora Transverberación con asistencia de los dos Trinidades y Santos San Elías San Simón San Ambrosio San Andrés Apóstol San Francisco de san Miguel San Ignacio de Loyola San Jerònimo San Juan Bautista decapitado San Nicolás de Tolentino San Pedro San Francisco Cristo de la Pasión con San Francisco San Pablo Santa Teresa herida con la flecha del Divino Amor María Magdalena San Agustín San Ambrosio San Jerónimo Santa Ana Martirio de una Santa La muerte de san Fransisco Javier (Xavier) Los Santos Ermitaños Pablo y Antonio en el Desierto Sta. Catalina de Alejandría discutiendo con los sabios El martirio de san Aproniano San Francisco de Borja San Ignacio escribe las Constituciones en la cueva de Manresa San Francisco en Extasis Santa Teresa San Francisco Santa Bárbara Santa Catalina de Alejandria Visita de santo Domingo de Siles a santo Domingo de la Calzada Apostolado San Juan Bautista Santa Lucía Santo Niño de la Guardia Transverberación de santa Teresa San Francisco Javier Santa Ana San Antonio de Padua San Pablo Santiago el Menor San Nicolás Obispo de Mira La Virgen con el Niño, San Francisco y Santa Clara San Mateo Santiago Apóstol San Bartolomé Santa Justa Santa Clara San Agustín San Francisco Imposicion de la casulla a San Ildefonso, con donantes La Sagrada Familia ALA BADO SEA ELSMOS SACRA MENTO San Yldefonso Virgen del Rosario con Santo Domingo y San Francisco San Vicente Ferrer Meditaciones de san Ignacio sobre el infierno San Ignacio escribe las Constituciones en la cueva de Manresa Ángel virtud Santo Tomás y el obispo Faustino de Padua San Ambrosio Arcángel San Miguel Jesús en casa de la MagadalenaCristo en casa de Marta y María Santo Domingo predica el rosario Ángel Virtud Arcángel San Rafael Aspiel Apetus Dei (Arcángel Arcabucero) Ángel de la Guarda Ángel de la pasión con escalera Arcángel San Miguel Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Virgen con el Niño Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe El milagro de la estampación de la imagen de la Virgen de Guadalupe en la tilma de Juan Diego Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Virgen con el niño, San José y San Roque Virgen con Niño Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe La Azucena de Quito Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Segunda y cuarta aparición de la Virgen de Guadalupe al indio Juan Diego Virgen del Rosario Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Tercera aparición de la Virgen de Guadalupe al mismo Virgen de la Antigua Copia de la Virgen de GuadalupeCopia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Primera aparición de la Virgen de Guadalupe al indio Juan Diego Segunda aparición de la Virgen de Guadalupe al mismo Virgen coronada por La Trinidad Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Milagro de la Virgen de Valvanera Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Copia de la Virgen de Guadalupe Primera y tercera aparición de la Virgen de Guadalupe al indio Juan Diego of baroque paintings proved to be a resilient one that allowed for the integration of many local aspects and techniques while keeping all those technical and thematic features that made a painting from Cusco (Peru) similar to one from Puebla (Mexico), Madrid (Spain), or Antwerp (Belgium). 4.2 Knowledge Influence: elBulli Graph This case proposes a formalization of the evolutionary method of creation developed by Ferran Adri` a in his restaurant, elBulli, over many years of work that took it to be considered the best restaurant in the world by expert media and the public. If a recipe is a sort of algorithm that explicitly gives a set of instructions to reproduce a given dish, then they are excellent for reproducing existing culinary knowledge. But if reproduction is not the objective any more, as Adri` a discovered early on, then the recipe would be just a byproduct of the whole method, not the corner stone. For the new method of creative cuisine developed by Adri` a, creativity was to be anchored at the levels of preparations, techniques and concepts as it is at these levels that creativity in the kitchen can become evolving creativity. Figure 5 shows the general schema of a recipe as conceived by Adri` a’s cooking style. Figure 5: Schema of a Recipe in elBulli context. What Adri` a and his team did was to develop and implement a method to achieve what Horng and Lin (Horng and Lin, 2009) have described as the objective of a truly creative form of culinary art, that is, a ”wider variety of dishes, as well as more innovative, aesthetically, and culturally innovative individual dishes”. The development of new techniques, the inclusion of new concepts, the constant addition of new products and, specifically, the combination, mix and merger of all elements would open the floodgates of discovery into the creation of new preparations and, along with it, a considerable increase in the rate of innovation. All these elements become nodes in a complex network that keep most of their own instances connected, a network that allows for the creation of many paths and the integration of new elements, specially products,techniques, and preparations. The Figure 4: Example of graph generated by SylvaDB, exported to GEXF format and visualized and styled in Gephi. This graph represents the similarity network of HispanicAmerican paintings in the period from 1600 to 1625. Colors designate modularity classes. network of paintings, and the role of art as an institution that contributes to sustain large-scale societies (Su´arez et al., 2012). From a set of 211 keys or descriptors it was carried out a manual semantic annotation of all artworks (with an average of 5.85 descriptors/work and peaks of 14 per work). This process of annotation required a special level of reliability in order to avoid data conflicts. A team of annotators was setup following the next hierarchy: administrators (with global permissions over the graph), reviewers (with all CRUD permissions over the graph data, but not over the schema), and annotators (with creation and edition permissions only). The fact that SylvaDB has a built-in support for collaborative work and detailed permissions management (both essential in this kind of research), was crucial to the success of the project. On the other hand, this experience allowed us to know how real users face the interface, resulting in a positive feedback about its usability and easiness when it comes time to handle complex data or modify the schema according to the necessities in real time. Altogether, more than 30 people were using SylvaDB in creating the Hispanic Baroque paintings network, and currently it is available for researchers all over the world to enhance and validate the content. The research focuses on the network of paintings resulting from an analysis of the edges that connected them (Su´arez et al., 2011) (see Figure 4). These edges are part of a detailed ontology that allowed the research team to fully categorize artworks from various provenances and geographic contexts. The network combinatory potential of the method is unquestionable, but even more important are the effects provoked by the integration of new elements connected in surprising ways with the existing structure. This increase in innovation can be seen in the growth in the numbers of hubs and relations in their method that would result in a multiplication of the dishes and recipes produced at the end of any given season. Using the SylvaDB query system, in this case running real Cypher queries for a Neo4j graph backend, conclusions about elBulli artistic evolution and the elements involved in its creative process, just came up. In this way, we could prove how useful is the representation of data in graphs, not only at conceptual level, but from the point of view of storage backend and querying. 4.3 Social Network Analysis: Preliminaries Project Finally, the Preliminaries project uses SylvaDB to store, visualize and analyze a large data set dealing with information contained in the 17th-century books that allow us to unearth the social network of publishers, writers, bureaucrats, and noblemen involved in the publishing industry at both sides of the Hispanic Atlantic. The ”preliminary” of a 17th-century book encompasses all pages appearing in the printed text before the beginning of the work itself. This information is divided into five different types: details of publication, documentation of censorship (both civil and ecclesiastical), licensing, selling price and dedications. The important thing for this project lies in the fact that all sections usually include names of the officers or individuals signing the specific documentation, dates and places of the issuing of the documents, i.e., in a few pages these preliminaries give us a very complete image of the individuals, places, dates, councils, and institutions involved in the publication of each specific title of literature. In this process, SylvaDB played an important role by allowing an incremental design of the schema without losing any previous data. By compiling all this information into a graph database, and subsequently performing different kinds of queries according to the various research questions, the dataset became a valuable source of information about the historical networks that influenced the publication of early Modern Spanish literature. This project has produced a research model that combines network-based analysis with quantitative and qualitative studies of early Modern cultural production. It has also explored the concept of cultural Figure 6: Screenshot of the SylvaDB schema for the Preliminaries Project graph. network within the framework of complexity theory and with new data covering various historical periods. Last, it has given evidence of the interaction between political structures and cultural production through social network analysis. The cultural objects considered in this study are complex and variable (see Figure 6). Therefore, the flexibility of SylvaDB, that allows for the adaptation of the schema in real time, was a key factor to achieve the goals of the project. 5 CONCLUSIONS AND FUTURE WORK SylvaDB, although still in private beta, is currently being used in several historical and cultural projects, and it will soon be open for any researcher or group to use. Also, it is open source software (GNU Affero General Public License18) and the source code can be easily found and forked19. As fields like Digital Humanities grow and evolve, more and more researchers with no training in computer science feel the need to work with large datasets and combine their traditional object-focused research with data analysis and visualization. SylvaDB aims at lowering the barrier for the adoption of database systems into the practices of those fields. Also, it is designed to support the storage and analysis required by large, advanced research projects looking for a powerful and intuitive data and document management system based on flexible schemas. As any other existing piece of software, SylvaDB is far of being finished in terms of features. A non exhaustive list of proposed features is shown below. 18http://www.gnu.org/licenses/agpl-3.0.html 19http://github.com/CulturePlex/Sylva •A visual editor for queries. Even if the natural language query input is powerful and intuitive enough, a more complete and customizable system that enables users to build their own complex queries is in our plans, due to the same criteria of usability and low barriers that got the project started in the first place. •Implementation of topological pattern matching protocols as a complement for the query module. •Better visualizations. Full-screen mode, integration with queries system, and basic interaction with properties of drawn elements such as shape, color, and size of nodes, relationships and labels. •A battery of well-known algorithms. Including the most common ones, like Page Rank, HITS or Dijkstra (Ding et al., 2002), could be a valuable help for users to perform first level analysis using the same interface. We are also working to improve tests coverage to reach at least a 90% of the code covered. ACKNOWLEDGEMENTS We acknowledge the support of the Social Sciences and Humanities Research Council of Canada through a Major Collaborative Research Initiative on The Hispanic Baroque. And the Canada Foundation for Innovation through the Leaders Opportunity Fund. REFERENCES Angles, R. and Gutierrez, C. (2008). Survey of graph database models. Computing Surveys, 40(1):1. Bastian, M., Heymann, S., and Jacomy, M. (2009). Gephi: An open source software for exploring and manipulating networks. Burza´ nska, M., Stencel, K., Suchomska, P., Szumowska, A., and Wi´ sniewski, P. (2010). Recursive queries using object relational mapping. Future Generation Information Technology, pages 42–50. Catarci, T., Costabile, M., Levialdi, S., and Batini, C. (1997). Visual query systems for databases: A survey. Journal of visual languages and computing, 8(2):215– 260. Ding, C., He, X., Husbands, P., Zha, H., and Simon, H. (2002). Pagerank, hits and a unified framework for link analysis. In Proceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval, pages 353–354. ACM. Ellis, G., Finlay, J., and Pollitt, A. (1994). Hibrowse for hotels: bridging the gap between user and system views of a database. In IDS’94 Workshop on User Interfaces to Databases, pages 45–58. Esterbrook, C. (2001). Using mix-ins with python. Linux Journal, 2001(84es):7. Fielding, R. (2000). Architectural styles and the design of network-based software architectures. PhD thesis, University of California. Fitzpatrick, B. (2004). Distributed caching with memcached. Linux journal, (124):72–74. Hacigumus, H., Iyer, B., and Mehrotra, S. (2002). Providing database as a service. In Data Engineering, 2002. Proceedings. 18th International Conference on, pages 29–38. IEEE. Hendrix, G., Sacerdoti, E., Sagalowicz, D., and Slocum, J. (1978). Developing a natural language interface to complex data. ACM Transactions on Database Systems (TODS), 3(2):105–147. Holovaty, A. and Kaplan-Moss, J. (2009). The Definitive Guide to Django: Web Development Done Right. Apress. Horng, J. and Lin, L. (2009). The development of a scale for evaluating creative culinary products. Creativity Research Journal, 21(1):54–63. Knoke, D., Yang, S., and Kuklinski, J. (2008). Social network analysis, volume 2. Sage Publications Los Angeles, CA. Malliga, P. (2012). Database services for cloud computing– an overview. Database, 2(3). Popescu, A., Etzioni, O., and Kautz, H. (2003). Towards a theory of natural language interfaces to databases. In Proceedings of the 8th international conference on Intelligent user interfaces, pages 327–327. ACM. Roddick, J. (1992). Schema evolution in database systems: an annotated bibliography. ACM sIGMOD record, 21(4):35–40. Santucci, G. and Palmisano, F. (1994). A dynamic formbased data visualiser for semantic query languages. Interfaces to Database Systems, pages 249–265. Su´ arez, J., Sancho, F., and de la Rosa, J. (2011). The artspace of a global community: the network of baroque paintings in hispanic-america. In Culture and Computing (Culture Computing), 2011 Second International Conference on, pages 45–50. IEEE. Su´ arez, J., Sancho, F., and de la Rosa, J. (2012). Sustaining a global community: Art and religion in the network of baroque hispanicamerican paintings. Leonardo, 45(3):281–281. Vicknair, C., Macias, M., Zhao, Z., Nan, X., Chen, Y., and Wilkins, D. (2010). A comparison of a graph database and a relational database: a data provenance perspective. In Proceedings of the 48th annual Southeast regional conference, page 42. ACM. Wolfe, R., Needels, M., Arias, T., and Joannopoulos, J. (1992). Visual revelations from silicon ab initio calculations. Computer Graphics and Applications, IEEE, 12(4):45–53. Zloof, M. (1975). Query by example. In Proceedings National Computer Conference, pages 431–438. ACM.