Last week I attended the 9th edition of iEMSs in Fort Collins, Denver. IEMSs is a bi-annual conference that brings together between 300 and 400 researchers from software engineering, intelligent systems, environmental modeling and decision making domains (among others). There were very few people that knew about ontologies and Semantic Web, which makes it a unique experience to learn about the problems from other communities. Going to this kind of events (outside of your community of expertise) has been eye opening for me in the past, and I cannot recommend it enough. Get out of your community bubble once in a while J
What was I doing at iEMSs?
I attended the conference to present 3 papers about our Model Integration project (MINT). The papers describe an overview of the project, in which we aim to reduce the time required to integrate together models from climate, hydrology, agriculture, economics and social sciences. In addition, we introduce a new approach to describe model variables and processes using the Ontosoft software registry and our plan to integrate Pegasus and Emely for efficient model coupling. More information is available in the conference program (hopefully our papers will soon be available in the conference proceedings as well). Overall, the presentations were well received and I was glad to learn that there is huge interest in some of the problems we are tackling, such as the description of models to facilitate their reusability or enabling model coupling.
One of the best parts of the conference were the keynotes. Temple Grandin started on Monday with a cry for acceptance of visual thinkers (“I see risk, other people try to measure it!”) together with the need to get closer to the infrastructure we use every day. Get out of the office and get your hands dirty once in a while!
The last keynote speaker was Thomas Vilsack, former US Secretary of Agriculture under the Obama administration. This is the first keynote I have seen given by a politician, with no slides and a direct but compelling speech. The speaker tackled several problems related to modeling, from the role of science in different debates (GMOs and climate change) to the need for new sustainable solutions given the increase of population around the globe. How can we make models that convince farmers and policy makers about the long term consequences of their actions? How can models be used to increase the productivity per individual acre? Can we find solutions so we become better consumers of food? How can we reduce and reuse food waste?
Excellent wake up call from former US Secretary of Agriculture, Thomas Vilsack on conflicting short- and -long term challenges in Agriculture- e.g. ecosystem markets #iEMSs2018pic.twitter.com/fjEa6r4V7m
Given that many sessions happened in parallel, this is a personal vision with the highlights of the talks I attended to:
Ibrahim Demir’s FloodAI is a very cool approach that mixes science with visual explanations early detection observations. They have done an impressive amount of work to be able to communicate their results with chat bots. No wonder why he won a conference award!
Alexei Voinov described surveys, tools and methods for participatory modeling. Remaining challenges are a) people tend to use the tools and models they are more familiar with, rather than experiment new ones in different contexts; b) Failure in method execution is not reported.
Ruth Falconer (University of Abertay) and the use of videogames in environmental modeling.
Sarah Mubareka’s report on integration of models of biomass supply. Creating accurate indicators for estimating biomass in Europe is a real challenge, as everyone one uses different definitions and metrics in their country.
Last week I attended the annual EarthCube All Hands Meeting (ECAHM) in Alexandria, Washington. Since it’s been a while since I last wrote my last post, I think it would be interesting to share my notes and highlights here for anyone who missed the event.
ECAHM meetings are usually very enriching experiences, as they bring together a variety of researchers from different fields related to geosciences, ranging from computer scientists to volcanologists or marine biologists. The purpose of the meeting is to gather the community together and hear everyone report back from their EarthCube NSF funded projects, which are targeted towards improving cyber-infrastructure in the geosciences. As a computer scientist, I think this is a great meeting to attend for two main reasons: first, you always learn something new, even if it’s not in your domain. Second, people are extremely grateful to your contributions, as you are helping them become more effective when doing their science.
So, what was I doing at ECAHM 2018?
I attended the meeting to present our latest progress in OntoSoft, a distributed software metadata registry we created at ISI to facilitate scientists describe their software. You can see the poster abstract online (and soon the poster itself). I also participated on a “speed-dating session”, where I got to discuss for half an hour how to describe software with a domain scientist; and I substituted Yolanda Gil in a panel for external partnership opportunities, where I presented the Open Knowledge Network initiative. This effort, led by NITRD, is a great opportunity of creating a shared open knowledge graph that would be used for both research and industry to refine and curate its contents. The idea is that this knowledge graph becomes part of the US infrastructure the same way supercomputers currently are, so anyone could benefit from it and also contribute to it. It looks like the NSF is keen to pursue this objective too.
Two colleagues of mine also presented other initiatives I am involved in. Deborah Khider showcased our efforts towards structuring metadata and creating standards in the paleoclimate sciences, together with a set of tools that a team of paleo-climate scientists have developed to work with that structured data. She also managed to mix Star Wars and Star Trek themes in her poster and presentation, which was well received by the attendants (I think everyone stopped at her poster)
As expected, keynotes at ECAHM are nothing like venues such as AAAI or IUI. The first speaker was Dean Pesnell (NASA) and he presented the research carried out by his team on studying the sun and sun spots. Why is this related to geosciences? Because the sun could be considered “our ground truth for the universe”, and anything related to its activity has many implications in any of the fields of geosciences. Their main problem is how to analyze the amount of data that they have. Each of their datasets may contain several hundred million images, so proper metadata is crucial (you don’t want to find out you have downloaded 300 million images for nothing). Dean showed some impressive videos of their observations of the sun, as well as their pipelines to handle “very big data” analyses.
The second speaker was Sarah Stamps, and she talked about continental rift and the Tanzania Volcano observatory. Apparently, geologists are one of the few people in the word who would run towards an erupting volcano, instead of away from it. Sarah described the EARS system (East African Rift System) they are setting up, and how they teamed up with CHORDS to enable real time analysis of the observations they measure on the field. Thanks to her work, they are developing an early warning system for hazard detection! Sarah was departing soon to set a few more observing stations in the field, so best of luck!!
The third speaker was Caroline S. Wagner, who gave some metrics on the social side of interdisciplinary collaboration across disciplines. Science has become increasingly collaborative and team based, and the number of international collaborations have doubled in the past years. The number of countries producing 95% of research has gone from 7 to 15, which indicates we are moving in the right direction. However, more than 50% of the articles are currently never cited. A few takeaways from this talk are: 1) International collaborations start face to face, so go to different events and meet new people; 2) Diverse teams usually take longer to be productive, as people don’t usually speak the same language. Be patient!!; 3) Work towards a solution, not towards interdisciplinar teams. Interdisciplinarity should be the means to an end, not the end itself.
Below are some additional highlights I found interesting for the EarthCube community.
Eva Zanzerika reported on the NSF 10 Big Ideas, which nicely summarize the interests of the agency in terms of funding in the next years. The report has been out since more than 1 year ago, but it’s never too late to catch up!
Doug Fils presented their plan for turning P418 turning into something bigger. In case you don’t know, P418 currently tracks the metadata of datasets exposed as schema.org and aggregates it in a search engine (a search engine for scientific data). Future plans are to ingest other types of resources and make the code base stable.
Interesting working lunch idea: A napkin drawing exercise. Do you know how to present your idea with a simple sketch?
After a few days back in Madrid, I have finally found some time to write about the eScience 2014 conference, which took place last week in Guarujá, Brasil. The conference lasted for 5 days (the first two days with workshops), and it got attendants from all over the world. It was especially good to see many young people who could attend thanks to the scholarships awarded by the conference, even when they were not presenting a paper. I found a bit unorthodox that the presenters couldn’t apply for these scholarships (I wanted to!), but I am glad to see this kind of giveaway. Conferences are expensive and I was able to have interesting discussions about my work thanks to this initiative. I think this is also a reflection of Jim Gray’s will: pushing science into the next generation.
We were placed in touristic resort in Guarujá, at the beach. This is what you could see when you got out of the hotel:
And the jungle was not far away either. After a 20 minute walk you were able to arrive at something like this…
…which is pretty amazing. However, the conference schedule was packed with interesting talks from 8:30 to 20:30 most of the days, and in general we were unable to do some sightseeing. In my opinion they could have reduced one workshop day and relax the schedule a little bit. Or at least remove the parallel sessions in the main conference. It always sucks to have to choose between two different interesting sessions. That said, I would like to congratulate everyone involved in the organization of the conference. They did an amazing job!
Another thing that surprised me is that I wasn’t expecting to see many Semantic Web people, since the ISWC Conference occurred at the same time in Italy, but I found quite a few. We are everywhere!
But let’s get back to the workshop, demos and conference. As I introduced above, the first 2 days included workshop talks, demos and tutorials. Here are my highlights:
Workshops and demos:
Microsoft is investing on scientific workflows!: I attended the Azure research training workshop, were Mateus Velloso introduced the Azure infrastructure for creating and setting up virtual machines, web services, webs and workflows. It is really impressive how easily you are able to create and run experiments with their infrastructure, although you are limited to their own library of software components (in this case, a machine learning library). If you want to add your own software, you have to expose it as a web service.
Impressive visualizations using Excel sheets at the Demofest! All the demos belonged to Microsoft (guess who was one of the main sponsors of the conference) although I have to admit that they looked pretty cool. I was impressed by two demos in particular, the Sanddance beta and the Worldwide Telescope. The former is used to load Excel files with large datasets to play with the data, select, filter and plot the resources by different facets. Easy to use and very fluid in the animations. The latter was similar to Google Maps, but you were able to load your excel dataset (more than 300K points at the same time) and show it on real time. For example, in the demo you could draw the itineraries of several whales in the sea at different points in time, and show their movement minute after minute.
New provenance use cases are always interesting. Dario Oliveira introduced their approach to extract biographic information from the Brazilian Historical Biographical Dictionary at the Digital Humanities Workshop. This included not only the life of the different persons collected as part of the dictionary, but also each reference that contributed to tell part of the story. Certainly a complex and interesting use case for provenance, which they are currently refining.
Paul Watson was awarded with the Jim Gray Award. In his keynote, he talked about the social exclusion and the effect of digital technologies. Having a lack of ability to log online may stop you from having access to many services, and ongoing work on helping people with accessibility problems (even through scientific workflows) was presented. Clouds play an important role too, as they have the potential for dealing with the fast growth of applications. However, the people who could benefit the most from the cloud often do not have the resources or skills to do so. He also described e-Science Central, a workflow system for easily creating workflows in your web browser, with provenance recording and exploring capabilities and the possibility to tune and improve the scalability of your workflows with the Azure infrastructure. The keynote ended by highlighting how important is to make things fun for the user (“gamification “ of evaluations, for example), and how important eScience is for computer science research: new challenges are continuously presented supported by real use cases in application domains with a lot of data behind.
I liked the three dreams for eScience of the “strategic importance of eScience” panel:
Find and support the misfits, by addressing those people with needs in escience.
Support cross domain overlap. Many communities base their work on the work made by other communities, although the collaboration rarely happens at the moment.
Cross domain collaboration.
Conference general highlights:
Great discussion in the “Going native Panel”, chaired by Tony Hey, with experts from chemistry, scientific workflows and ornithology (talk about domain diversity). They analyzed the key elements of a successful collaboration, explaining how in their different projects they have a wide range of collaborators. It is crucial to have passionate people, who don’t lose the inertia after the grant from the project has been obtained. For example, one of the best databases for accessing chemicals descriptions on the UK came out from a personal project initiated by a minority. In general, people like to consume curated data, but very few are willing to contribute. In the end what people want is to have impact. Showing relevance and impact (or reputation, altmetrics, etc.) will grant additional collaborators. Finally, the issue of data interoperability between different communities was brought up for discussion. Data without methods is in many cases not very useful, which encourages part of the work I’ve been doing during the last years.
Awesome keynotes!! The one I liked the most was given by Noshir Contractor, who talked about “Grand Societal Challenges”. The keynote was basically about how to assemble a “dream team” of people for delivering a product/proposal, and all the analyses that had been done to determine which factors are the most influential. He started by talking about the Watson team, who built a machine capable of beating a human on TV, and continued by presenting the tendencies people have when selecting people for their own teams. He also presented a very interesting study of videogames as “leadership online labs”. In videogames very heterogeneous people meet, and they have to collaborate in groups in order to be successful. The takeaway conclusion was that diversity in a group can be very successful, but it is also very risky and often it ends in a failure. That is why people tend to collaborate with people they have already collaborated with when writing a proposal.
The keynote by Kathleen R. McKeown was also amazing. She presented a high level overview of the work in NLP developed in their group concerning summarization of news, journal articles, blog posts, and even novels! (which IMO has a lot of merit without going into the detail). She presented co-reference detection of events, temporal summarization, sub-event identification and analysis of conversations in literature, depending on the type of text being addressed. Semantics can make a difference!
New workflow systems: I think I haven’t seen an eScience conference without new workflow systems being presented 😀 In this case the focus was more on the efficient execution and distribution of the resources. Dispel4py and Tigres workflow systems were introduced for scientists working in Python.
Cross domain workflows and scientific gateways:
Antonella Galizia presented the DRIHM infrastructure to set up Hydro-Meteorological experiments in minutes. Impressive, as they had to integrate models for meteorology, hydrology, pluviology and hydraulic systems, while reusing existent OGC standards and developing a gateway for citizen scientists. A powerful approach, as they were able to do flooding predictions on in certain parts of Italy. According to Antonella, one of the biggest challenges on achieving their results was to create a common vocabulary which could be understood by all the scientists involved. Once again we come back to semantics…
Rosa Filgueira presented another gateway, but for vulcanologists and rock physicists. Scientists often have problems to share data among different disciplines, even if they belong to the same domain (geology in this case). This is because every lab often records their data in a different way.
Finally, Silvia Olabarriaga gave an interesting talk about workflow management in astrophysics, heliophysics and biomedicine, distinguishing the conceptual level (user in the science gateway), abstract level (scientific workflow) and concrete level (how the workflow is finally executed on an infrastructure), and how to capture provenance at these different granularities.
Other more specific work that I liked:
A tool for understanding the copyright in science, presented by Richard Hoskings. A plethora of different licenses coexist in the Linked Open Data, and it is often difficult to understand how one can use the different resources exposed in the Web. This tool helps on guiding the user about the possible consequences of using a given resource or another in their applications. Very useful to detect any incompatibility on your application!
An interesting workflow similarity approach by Johannes Starlinger, which improves the current state of the art by making efficient matching on workflows. Johannes said they would release a new search engine soon, so I look forward to analyzing their results. They have published a corpus of similar workflows here.
Context of scientific experiments: Rudolf Mayer presented the work made on the Timbus project to capture the context of scientific workflows. This includes their dependencies, methods and data under a very fine granularity. Definitely related to Research Objects!
An agile annotation of scientific texts to identify and link biomedical entities by Marcus Silva, with the particularity of being capable of loading very large ontologies to do the matching.
Workflow ecosystems in Pegasus: Ewa Deelman presented a set of combinable tools for Pegasus able to archive, distribute simulate and re-compute efficiently workflows. All tested with a huge workflow in astronomy.
Provenance is still playing an important role in the conference, with a whole session for related papers. PROV is being reused and extended in different domains, but I still have to see an interoperable use across different domains to show its full potential.
In summary, I think the conference has been a very positive experience and definitely worth the trip. It is very encouraging to see that collaborations among different communities are really happening thanks to the infrastructure being developed on eScience, although there are still many challenges to address. I think we will see more and more cross domain workflows and workflow ecosystems in the next years, and I hope to be able to contribute with my research.
I also got plenty of new references to add to the state of the art of my thesis, so I think that I also did a good job by talking to people and letting others know of my work. Unfortunately my return flight was delayed and I missed my connection back to Spain, converting my 14 hour flight home to almost 48 hours. Certainly the longest journey from any conference I have assisted to.
While being a PhD student, many people have asked me about the subject of my thesis and the main ideas behind my research. As a student you always think you have very clear what you are doing, at least until you have to actually explain it to someone who is not related to your domain. In fact, it is about using the right terminology. If you say something like “Oh yeah, I am trying to detect abstractions on scientific workflows semi-automatically in order to understand how they can better be reused and related to each other”, people will look at you as if you didn’t belong to this planet. Instead, something like “detecting commonalities in scientific experiments in order to study how we can understand them better” might be more appropriate.
But last week the challenge was slightly different. I was invited to give an overview talk about the work I have been doing as a PhD student. And that is not only what I am doing, but why am I doing it and how is it all related without going into the details of every step. It may appear as an easy task, but it kept me thinking more than I expected.
The last 3 years I have been involved in the Wf4Ever project, which has developed the notion of Research Objects and their respective models (previously introduced another post). Lately I have been exploring new ways for eating my own dog food by associating Research Objects to my papers as HTML web pages (see an example here). These Research Objects are useful, as they serve as summary for the paper in question, and they have pointers to all the datasets, queries and additional materials that I could not include in the paper.
However, I realized that I spent a lot of time creating them and annotating them. Therefore during last Christmas I have created a Research Object Creator tool, which takes as input a LaTeX file and extracts its title and abstract to create an annotated page in rdf-a. It also produces a structure of the contents to reference, so you only have to fill in (and annotate if you want) the resources to point to. A sample can be seen in the image below:
Thispart of the tutorial explains how to design a human readable documentation. When browsing an ontology, it is very important to provide accurate definitions and examples of how to use it. If these are not provided, the ontology will be very difficult to reuse. Having a documentation easy to navigate, which explains every concept and relationship separately and which presents an overview and examples improves the understandability of the whole ontology to other people.
Some people address this step by pointing to a report/deliverable/paper where the ontology is described. Although this helps, it is not easy to navigate and will drive crazy any final user. I don’t recommend it. Furthermore, according to my experience, if the ontology is documented in a paper then the information will be of little use.
Making a proper documentation is difficult and takes time. Fortunately, there are some tools to help you overcome this task, like LODE, Parrot, OWLDoc, neologism, Ontospec, etc. I have worked with LODE, Parrot and OWLDoc, so I will only cover these here:
LODE: For me it’s the best of the tools I’ve tried. It is a web service that takes as input an owl file and generates an html. The html is W3C-style with the definition of each of the terms extracted from the domain, ranges and metadata of your owl file. If you extract the appropriate bits you can automatically create templates to customize your documentation with additional images, explanations and examples (like this one).
Parrot: Very similar to LODE, although the styles used are different and you have to clean some of the properties not defined within the namespace of your ontology (like the ones used to add metadata). It works really well, and my choice picking LODE instead of Parrot is a matter of styles.
OWLDoc: NeOn Toolkit plug-in that generates an owl documentation javadoc style from your .owl file. I don’t personally like it much, as customizing it is a bit of a pain.
Once you have your html template from one of these tools (with all the concepts of the ontology fully covered), you should add sections describing an overview of the model and examples. My suggestion is to follow the structure of W3C documents, namely:
Title and date of the release.
Metadata: Authors, contributors, version, imported ontologies, license, link to previous version, link to the latest version.
Abstract: small summary of your ontology in 2 lines. I recommend pointing to the owl file here as well.
Table of contents of your html document.
Introduction: provide context to the ontology. What are its goals and the benefits of using it?.
Namespace declarations: Namespace URIs of all the vocabularies used within the document (this could be found at the end as well).
Overview of classes and properties: Very small section with the list of tables and properties of the ontology, for making the navigation easier to the reader.
Description: Diagram of the ontology concepts, relationships and how they are related to each other. Usage examples might help clarifying things as well.
Cross reference section: this is the section automatically generated by the tools covered above. Just copy what they generated J
Acknowledgements, specially remember to include the developers of the tools you have used.
Want to see some examples? Check PROV (W3C), some commonly used vocabularies like foaf or Dublin Core (which cover the points listed above with their own structure) or some of the ontologies I’ve been publishing, like p-plan or wf-motifs. Note that the order in which the points of the list appear is not mandatory. Modify it in order to make your ontology easier to use to the final user!
A scientific workflow can be defined as the computational steps required for executing an in silico experiment. Scientific workflows are similar to laboratory protocols. The only difference between them is that in scientific workflows scientists run their simulations with a computer instead of in a laboratory. These kinds of workflows are becoming increasingly important for three reasons:
They help the scientist researching the experiment to repeat the results and convince his/her colleagues of the validity of the method followed.
They expose the inputs, intermediate results, outputs and codes of the experiment. In this regard they can be considered the log of the research, which is very useful for the reviewers of the work.
They allow reproducing (and reusing) the method of the experiment to other scientists, since they just have to rerun the original experiments or use other input data.
On the left of this post there is an example of a workflow for feature selection: each of the steps represents a computational method that performs some operation on the input data received from the previous step. The initial input is a Dataset and a set of words, and the final output the selected features.
Repositories of workflows like myExperiment, CrowdLabs and Galaxy have been created for scientists to share their scientific workflows, so a reasonable question for a scientist would be: how do these workflows relate to each other? Which are the most popular workflow fragments in the dataset? Has there really been reuse among the different workflows?
A possible way to find out these popular workflow fragments is to apply graph matching techniques. If we represent the workflow as a directed acyclic graph, we can apply existing methods to see how a set of workflows overlaps with other similar workflows in the repository. In the UPM we have been recently working on this problem, and the technique we have used is described on our last accepted work at K-CAP. The idea is to obtain the best context free grammar encoding all the workflows of the repository. The rules of the grammar will be the most common fragments among the workflows, which is exactly what we are looking for. If you want more information and examples, I encourage you to have a look at the paper.