Aemoo: Linked Data exploration based on Knowledge Patterns

Tracking #: 1208-2420

Andrea Giovanni Nuzzolese
Valentina Presutti
Aldo Gangemi
Silvio Peroni
Paolo Ciancarini

Responsible editor: 
Guest editors linked data visualization

Submission type: 
Full Paper
This paper presents a novel approach to Linked Data exploration that uses Encyclopedic Knowledge Patterns (EKPs) as relevance criteria for selecting, organising, and visualising knowledge. EKP are discovered by mining the linking structure of Wikipedia and evaluated by means of a user-based study, which shows that they are cognitively sound as models for building entity summarisations. We implemented a tool named Aemoo that supports EKP-driven knowledge exploration and integrates data coming from heterogeneous resources, namely static and dynamic knowledge as well as text and Linked Data. Aemoo is evaluated by means of controlled, task-driven user experiments in order to assess its usability, and ability to provide relevant and serendipitous information as compared to two existing tools: Google and RelFinder.
Full PDF Version: 


Solicited Reviews:
Click to Expand/Collapse
Review #1
By Aba-Sah Dadzie submitted on 30/Oct/2015
Review Comment:

The authors have, overall, addressed my previous comments. The introductory sections and the background justifying the work and describing the interface are much easier to follow. The limitations are also more clearly explained. I’d recommend an accept subject to addressing the following.

One claim that probably needs toning down still is wrt to learnability and usability - 2nd para on p.21 - I can see only “0 (Hard to learn/use), 1 (Easy to learn/use)” in addition to the SUS. Both are a lot more complex than that question, and typically are derived based on a set of questions addressing specifically identified potential usability issues. Of course, that may have been done here, in which case I either missed it or the information is not provided. If the former - cross-reference here, if the latter please provide it.

Similarly, on p.23 - the authors hypothesise about support for relationship finding and also why google outperformed the visualisation tools. Did you by any chance go back to ask the participants for their opinions?

Re - literature review - I asked myself a couple of times - why refer to specific tools and then dismiss them as, e.g., “… they are not directly comparable with Aemoo as they …”

An interesting point I did not pick up on earlier: Google performed better at the RELATIONSHIP task than the two visualisation-based tools with a key function being relationship identification/visualisation! Google doing better is explained away by popularity - this may be so, am not debating that here - but might be worth finding out from the participants what may have contributed to this.

********* other minor points

p.2 - don’t understand the first half of this sentence -
“If Wikipedia article writing and hyperlinking confirm this intuition for each entity type in its ontology is an empirical matter, and we have performed data mining and experiments [37] to this aim, proving that automatically extracted EKPs can be trusted as relevant con- figurations of data for a given entity type.”

Fig4a - caption and main text refer to Alan Turing - actual diagram is about Kant

Is the relationship “between Prussia and Baltic See (cf. 4(b))” really unexpected - I would’ve thought it was obvious.

p.15 - “(i) although Google does not provide an interface specially designed for exploratory search, it is currently the most used exploratory tool on the Web. Users have developed their own methods for exploring and discovering knowledge by using Google, and they are very familiar with its interface. ” - sufficiently strong claim, especially since used later on to bolster the argument about usability of Aemoo, that it really needs some backing evidence.

Fig. 7 - this is is for the Likert scale (1-5 ?), right? Q2 & 3 say “I am an expert user of Aemoo/RelFinder” - and
“ They declared to have … no knowledge about RelFinder and Aemoo “ - but both bars have average above 1!

* a few typos/errors

Review #2
By Jan Polowinski submitted on 14/Nov/2015
Review Comment:

The issues concerning contradicting values for path popularity and the misleading phrases in the contribution have been fixed.

The minor issues that I remarked in the last review have been addressed carefully. There are still a few overfull lines (usually where typewriter style is used - I once found a fix for this, but I also have to google it. You may have to use manual hyphens here).

Also there are some, probably new AE/BE problems like ..zation vs. ..sation. I do not consider these things critical, though, since they don't actually harm the readability.

Review #3
By Mariano Rico submitted on 17/Dec/2015
Minor Revision
Review Comment:

Although authors say in the cover letter "Our answer: We removed the first part of hypothesis (i.e., the cognitive soundness of EKPs).", I see in the current version that:
1) section 4.1, entitles "Cognitive soundness of EKPs", and focuses on the experiment and results described in [37]. Your hypothesis 1 ("EKPs provide an unifying view as well as a relevance criterion for building entity-centric summaries) remains no evaluated.
2) section conclusions, says "Our work grounds on two working hypotheses: (i) the cognitive soundness of EPPs and (ii)...". You should change it according to the new hypothesis.

There is a "blue" in pag 13. You said that "Hence, we remove the term “blue” and we used “arrow” only.". in this case, "A blue link located" --> "A link located".

The rest of my comments have been correctly clarified and/or addressed.