Some research questions interested me

1. What're the differences between Intenet of Things and Semantic Web? Are they complements, overlapping or identical ?

2. In a stream reasoning scenario, why continous queries can perform reason? Isn't it reasoners do such a thing ?

- Incremntal maintenance of ontological entailments algorithm[1] does reasoning on transitivity of property in the window. However, it doesn't mention any reasoners invloved during the maintenance process. But I highly suspect that there is a reasoner that lives in the window. 

3. WHY RDF ?

 - The RDF model is based on a traditional persisted-data paradigm, where the focus is on maintaining a bounded set of data items in a knowledge base.[2] 

4. how to realize incremental reasoning ? 

- my question is, is incremental reasoning in stream reasoning system automatic? Is it true that as long as the reasoner is living in the stream reasoning system, the reasoning will be executed always, so that whenever new data arrives, reasoners are constantly providing entailments ? 

5. Why SPARQL is not suitable for RDF streams ?

- Some operators in SPARQL need the whole dataset to process, like distinct, join or sort .

6. How are the reasoners deployed to execute reasoning in parallel ? 

- The authors in this paper[3] proposed a scalable and efficient RDFS reasoning approach by splitting triples according to their predicates. They deploy one Triple-Splitter to distribute triples into three process elements(PE): DomainRange PE, SubClassPE and SubPropertyPE. Each of the PEs can perform RDFS reasoning under certain RDFS rulesets[4]. Since all these tasks are in parallel, the performance can be improved. 

 7. What's the relationship between Machine Learning and Semantic Web? Which is better in terms of what aspects ? 

- A blog[5] points out the challenges of combing machine learning and semantic web, with the statement saying that the ontology merging and maintenance is the way to go. Semantic Web tries to extend the current web by annotating the web of data, to create a "Meta-Web" that is rich in semantics. The ontology plays a very important role to realize data interoperability. Thus, machine learning provides a potential way for us to merge ontologies. 

8. Why RDF needs temporal model? 

RDF needs temporal model to (1) keep historical data (2) indicate the change of status. An example is illustrated in [6]: If John is an undergrate, and then he attends masters. The triple <john a undergraduate> will be break since this statement is no longer valid. <john a master> statement will be formed. Thus the graph loses the history data of "john was an undergraduate" and cannot answer the query of "Was john a undergraduate?". We need a temporal model attached with RDF model to be able to retrieve such time-related information.

Adding time element to RDF has two methods: (1) labeling (2) versioning. Labeling is preferred way because (1) it preserves the spirit of distributed and extensible nature of RDF (2) reduce the overhead in the scenarios where only a little data is updated. (otherwise, creating one version of the graph where little data is frequently changing is time-consuming).  

9. How to realize a semantic reasoner with a programming language ?

- How does Pellet work?

10. understand linked data by examples

- putting the data online is not enough. Publishers need to ensure that data is made available in both human- and machine-understandable formats and is linked to other data.[7]

- Linked Data, a set of four design principles put forward by Tim Berners-Lee in 2006, serve exactly that purpose. Hence, in order to publish Linked Data, publishers should:

1. Use Uniform Resource Identifiers (URIs) as names for things, e.g.  http://dbpedia.org/resource/Brussels can be used for referring to the city of Brussels.
2. Use HTTP URIs, so that people can look up those names.
3. When someone looks up a URI, provide useful information, using the standards (i.e. RDF, SPARQL).
4. Include links to other URIs, so more things can be discovered, e.g. fromhttp://dbpedia.org/resource/Brussels a link is available to http://dbpedia.org/resource/Belgium.

 11. What's SPARQL algebra?

 

--------------------------------------------------------------------------------------------

REFERENCE:

[1] Barbieri, Davide Francesco, et al. Incremental reasoning on streams and rich background knowledge. Springer Berlin Heidelberg, 2010.

[2] https://www.w3.org/community/rsp/rsp-workshop-2015/

[3] Hoeksema, Jesper, and Spyros Kotoulas. "High-performance distributed stream reasoning using s4."Ordring Workshop at ISWC. 2011.

[4] http://www.w3.org/TR/2014/REC-rdf11-mt-20140225/#rdfs-interpretations -> Section 9.2.1 Patterns of RDFS entailment 

[5] http://www.academia.edu/2542968/A_machine_learning_perspective_for_the_Semantic_Web

[6] Gutierrez, Claudio, Carlos Hurtado, and Alejandro Vaisman. "Temporal rdf." The Semantic Web: Research and Applications. Springer Berlin Heidelberg, 2005. 93-107.

[7] https://joinup.ec.europa.eu/community/semic/news/understanding-linked-data-example