RDF Presentation and Correct Content Conveyance for Legacy Services and the Web of Things Maxime Lefrançois Mines Saint-Étienne, Univ Lyon, Univ Jean Monnet, IOGS, CNRS, UMR 5516, LHC, Institut Henri Fayol F - 42023, Saint-Étienne, France [email protected] ABSTRACT 1 INTRODUCTION RDF aims at being the universal abstract data model for structured The Web today sees companies and web services exchanging data data on the Web. However, the vast majority of web services con- in various and multiple formats. XML (not RDF/XML) is still widely sume and expose non-RDF data, and it is unlikely that all these used, CSV is commonly used on Open data portals, APIs often services be converted to RDF one day. This is especially true for sen- use JSON. Constrained devices on the Web of Things [15, 16, 34] sors and other devices in the Web of Things, as most RDF formats tend to use concise formats, potentially binary such as EXI [29] or are verbose while constrained devices prefer to consume and ex- CBOR [5]. This raises the problem of semantic interoperability at pose data in concise formats. In this paper, we propose an approach the data level, i.e., how services and devices can get the meaning to make these services and things reach semantic interoperability, of the messages they exchange. RDF has the potential of being the while letting them the freedom to use their preferred formats. Our universal abstract data model for structured data on the Web, which approach is rooted in the Web’s architectural principles and the would be a first step towards enabling semantic interoperability on linked data principles, and relies on the definition of RDF presenta- the Web. Yet, RDF data formats (RDF/XML, Turtle, JSON-LD) will tions, which describe the link between RDF graphs and their rep- probably never be the only formats used on the Web. However, the resentations. We introduce the RDF Presentation ontology (RDFP) RDF data model can still be of use as lingua franca for semantic that can be used to model inputs and outputs of procedures of the interoperability. new Semantic Sensor Network ontology (SOSA/SSN), and input- In this paper we are investigating the issue of lowering the Data and outputData of interaction patterns of things in the W3C required effort for different services and devices to reach seman- WoT Thing Description ontology. We then propose practical solu- tic interoperability at the data level using Semantic Web models, tions for web agents to be able to discover how a message content without being too prescriptive on the format they should use. In can be interpreted as RDF, generated from RDF, or validated, with particular, we want to avoid complex compromises such as trying different Web interaction protocols. to stick to JSON-LD, and maintain an ontology, a JSON-LD context, and a JSON schema in parallel. We also want to avoid the need for CCS CONCEPTS using RDF syntaxes, however compact they may be such HDTQ or ERI [11]. • Information systems → World Wide Web; Studying semantic interoperability at the data level on the Web obviously involves some modeling of the communication between KEYWORDS heterogeneous agents on the Web. As a first approximation, let us Semantic Interoperability, RDF, RDF Presentation, RDF Lifting, RDF describe such communication as follows: a sender wants to send lowering, RDF validation some content (e.g., the state of a resource) to a receiver. It first gen- erates a representation of that content, transmits the so-formed ACM Reference Format: message via the Web to the receiver, that then decodes it and gets Maxime Lefrançois. 2018. RDF Presentation and Correct Content Con- some understanding of the original content. For this decoded mes- veyance for Legacy Services and the Web of Things. In 8th International sage to be equivalent to the original content, (1) all of the essential Conference on the Internet of Things (IoT 2018) (IOT ’18), October 15–18, characteristics of the content must be encoded in the message, (2) 2018, Santa Barbara, CA, USA. ACM, New York, NY, USA, 8 pages. https: the encoding and the decoding phase must be symmetric, and (3) //doi.org/10.1145/3277593.3277618 the message must not be altered in the transmission medium which is the Web. We name this Correct Content Conveyance (C3). We commit ourselves to develop an approach that conforms to, Permission to make digital or hard copies of all or part of this work for personal or and leverages, the Web’s architectural principles and the linked classroom use is granted without fee provided that copies are not made or distributed data principles. As Web’s foundations can be considered domain for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the knowledge like any other, we propose to adopt a rigorous knowl- author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or edge engineering methodology to answer our research question. republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]. We therefore apply the following steps, according to which this IOT ’18, October 15–18, 2018, Santa Barbara, CA, USA paper is structured. © 2018 Copyright held by the owner/author(s). Publication rights licensed to Associa- Step 1, analyze the domain. Section 2 formalizes the con- tion for Computing Machinery. ACM ISBN 978-1-4503-6564-2/18/10...$15.00 cepts of RDF Presentation, Lifting, Lowering, and Validation, https://doi.org/10.1145/3277593.3277618 which are used in the rest of this paper. IOT ’18, October 15–18, 2018, Santa Barbara, CA, USA Maxime Lefrançois Step 2, develop scenarios. Section 3 depicts three scenarios stream into an RDF graph, and the process that consists of using where agents communicate on the Web, all of which involve an RDF presentation to encode an RDF graph into a typed octet some adaptation of one or the other party to enable Correct stream, respectively.1 We formalize these terms as follows. Content Conveyance. Definition 2.2 (Lifting Rule). R" : S!G is a lifting rule for p if Step 3, extract competency questions. Section 4 lists com- petency questions that we derive from the scenario to define ¹ д;sº»hд;si 2 p ) R"¹sº = д¼ the scope of the ontology. 8 Definition 2.3 (Lowering Rule). R# : G!S is a lowering rule for Step 5, develop the ontology. Section 5 provides ontolog- p if ical representation for the core concepts formally defined in Step 1, that gravitate around the concept of RDF presentation. ¹ д 2 Gº ¹ s 2 Sº»hд;si 2 p¼ ) 8 9 The result of this step is the RDF Presentation (RDFP) ontol- # ¹ s0 2 Sº»hд;s0i 2 p and R ¹дº = s0¼ ogy, along with simple alignments to the Open Geospatial 9 Consortium (OGC) and World Wide Web Consortium (W3C) Definition 2.4 (Valid RDF Graph). Let p 2 P be an RDF presenta- Semantic Sensor Network ontology (SOSA/SSN) [17], and to tion. An RDF graph д is valid for p iff the W3C Thing Description ontology (TD) [20] that show s 2 S s.t. hд;si 2 p how it can be used in practice. RDFP is the main contribution 9 of this paper, and is used in the SOSA/SSN specification. The validation rule of p is an application from G to ftrue; f alseg, Step 6, qualitatively validate the ontology by showing that associates each graph д 2 G with the value true if and only if how it answers the competency questions and enable д is valid for p. the scenarios. This step is partly led throughout Section 5, Definition 2.5 (Valid Representation). A typed octet stream s is a and Section 6 shows how the RDF Presentation ontology and valid representation for p iff the concepts it defines can be used with some of the Web interaction protocols to answer the complementary set of д 2 G s.t. hд;si 2 p 9 competency questions from Section 4, and enable scenarios The representation validation rule of p is the application of S to from Section 3. Namely, how can a web agent discover the ftrue; f alseg, that associates each typed octet stream s 2 S with information about the RDF presentation that is used by its the value true if and only if s is a valid representation for p. interlocutor, either directly, or indirectly. As illustrated on Figure 1, many formalisms exist to specify lift- 2 PARTIAL FORMALIZATION OF THE ing rules. We can cite the SPARQL-Generate rules [24, 25], the RML DOMAIN mapping language [10], XSPARQL rules [1], the CSV metadata [33] and GRDDL [7].2 Lowering rules can be specified using STTL [8] In this section, we formalize the concept of RDF presentation, in or XSPARQL [1]. Then, the SPIN formalisms [22], Shape expres- accordance with the definitions from [18], [28] and [9], that define sions [14, 27], and W3C SHACL, can be used to define validation the fundamental principles of the Web and of the Web of Data. rules for RDF graphs. As for formalisms that can be used to define We also propose definitions for three additional concepts: RDF representation validation rules, every XML-based Internet media presentation, RDF lifting, and RDF lowering. Figure 1 illustrates a types can be validated using XML Schema, while JSON Schema can specific RDF Presentation, and lists some of the existing formalisms validate those based on JSON.