Talks

About

This space is for collection of material used in presentations about the project.

See also BackgroundMaterial.

Demo No.1 (May 2008)

Rough outline of the Demo that Travis gives of INSPIRE. Uses the INSPIRE alpha site. I do not post live links below, but all search strings can be entered in the main search box at hep-inspireDOTnet or, for the first part, at http://www.slac.stanford.edu/spires/

SPIRES Strengths: What we must reproduce

• Powerful, Flexible Query Language
  • find title higgs boson and a haber, h and k supersymmetry
• Intuitive (mostly) powerful author name handling
  • find author l j dixon
• Citation analysis
  • find a Dixon, l j citesummary (i.e. select citesummary from the pull-down menu at top, and click display again)
  • fin a ellis, john citesummary (from above page, just change the search to find a ellis, john...will take forever)
  • find aff SLAC citesummary (change search to find aff SLAC...will time out and give a blank page)

SPIRES Weaknesses: What we hope NOT to reproduce

• Speed/Power (as we just saw)
• Query language is a barrier to entry
  • type quark ellis in search box, see what happens...
• ASCII only
  • find t effective higgs and a toscano
• Bloated, complex codebase

INSPIRE Demo System "alpha" (go to th alpha system and use that for remaining demo)

• SPIRES Query Language
  • find a peskin, m (click around from there...)
• Brief format comforting, few differences
• Detailed format, expanded presentation of all information
• Transitional interlinking with SPIRES
• SPIRES data already there

INSPIRE Improvements: Simpler query alternative

• Suggest other searches
  • author:weinstien (note that it suggests corrections)
• google like
  • scalar peskin chou
• references tab (click on "detailed record", and then references tab)

INSPIRE Improvements: Native utf-8 and i18n, even TeX

• examples of display
  • yusuke nishida
  • effective higgs toscano
  • scalar peskin chou

INSPIRE Improvements:Beginnings of interactivity social features

• Will fit into existing community infrastructure
• Login
• Discussion tab/ratings

INSPIRE Improvements:Authors

• find a dixon, l j then click on highlighted name in first record
• Summary of information about an author history of affiliations, co-authors (needs tweaking) and citing

INSPIRE Improvements:Citations, Citations, Citations

• Citation demo
  • find a dixon, l j
    • rank by citation
    • detailed tab-> citations
    • Co-citations
• Citation Analysis/Summary
  • find a ellis, john
  • find a SLAC (as before, but works, and fast!)
  • (i.e. enter nothing in the search box and hit return...takes a while, but does the full database!) * Speed! * Capability to do large sets

INSPIRE: Next Steps

• Clean up few small issues with alpha interface
• Build data curation tools
  • SPIRES currently has excellent data due to feeds and harvesting...
  • ...but also much maintenance
• Curation tools must be more interactive and must leverage:
  • SPIRES+CDS expert staff
  • HEP Community (via interactive applications with the users- tagging etc.)

Demo No.2 (May 2009)

Demo given by Tibor at the 3rd HEP Information Resource Summit, Fermilab, 2009-05-06. Demoing mostly progress since the 2nd Summit. Some user-level features (plots, keywords, autocompletion); some admin-level features (record editing, merging).

1. Demo of plots

Story: I recall seeing a cool plot about hadron production cross section from Beijing collaboration. Search for hadron cross section. Too many results. Refine by adding bes and slac since it occurs to me they were published as SLAC reports. So the search string becomes hadron cross section bes slac. Three results. See the plots. Click on second hit, last plot, to see legend. Note display of TeX formulas. Click on image to get closer view. Muse about points behind the plots. Muse about automatic extraction from source tarballs.

URLs in this demo:

• http://inspiredev.cern.ch/search?p=hadron+cross+section+bes+slac
• http://inspiredev.cern.ch/record/542796#plots
• 

2. Demo of root files

Story: we would like to archive work going into an MSc thesis. ROOT as a de facto standard for visualization and analysis of large amounts of data in HEP. Example: MSc thesis of Barbara Storaci. Search for storaci. Choose thesis record. Click on Fulltext tab. Note PDF and some ROOT files. Note the size 700MB. Muse about storage requirements. Use smaller files for demo of remote access to files. Run ROOT at laptop like this:

   $ root
   root [0] TFile *f = TFile::Open("http://inspiredev.cern.ch/record/752281/files/mlpHiggs.root")
   root [1] b = TBrowser()

and go to the ROOT Object Browser window, click on `ROOT Files', then on `http://inspiredev.cern.ch/record/752281/files/mlpHiggs.root', then on some dir (e.g. bg_filtered) and some graph (e.g. acolin).

URLs in this demo:

• http://inspiredev.cern.ch/search?p=storaci
• http://inspiredev.cern.ch/record/752281/files/

2.5 Mathematica Files

I read a paper by Lance Dixon about ultraviolet behavior of supergravity and it had some extra fiels attached but I couldn't find them in arXiv.

[See here: http://arxiv.org/abs/0905.2326] [TODO: put a link to arXiv from the inspire record...]

Muse about the note in the ocmments about windows incompatibility, the difficulty of finding and unpacking the tarball and the impermanence of "this http URL"

Now search inspire:

dixon ultraviolet supergravity

Take first one, click on detailed->fulltext

see the full set of extra mathematica files and pdfs available. If you have mathematica reader (free) installed and configured on your demo computer, click the .nb files and mathematica will open, you can rotate 3D diagrams of the integrals in the paper. Otherwise the pdf will give a sense of what is happening.

URLS:

• http://arxiv.org/abs/0905.2326
• http://inspiredev.cern.ch/record/772790/files/

3. Demo of keyword clouds

Demo in CLI of keyword extraction for a recent nice record:

   $ /opt/cds-invenio/bin/bibclassify -k HEP.rdf ~/private/tmp/0903.3933v1.pdf

Note various sections of the output. The keyword classifier is usually run in the background for the inputters to pick good candidates from. Can be used for keywords clouds and clustering later. Taxonomy vs folksonomy. Example: search for jongwook. Choose the first m-theory paper, 0804.3349. See attached fulltext. Click on Keywords tab and Generate keywords button. See the cloud. (Beware, links in the cloud do not work, plus could has some unwanted standalone kws.)

URLs in this demo:

• http://inspiredev.cern.ch/search?p=jongwook
• http://inspiredev.cern.ch/record/764167/keywords

4. Demo of autocompletion

Example of AJAX tools. HEPnames. Submit a new John Doe entry. Check out autocompletion on the thesis advisor field. Speak of community-organized maintenance and curation.

URLs in this demo:

• http://www.slac.stanford.edu/spires/hepnames/additionsNEW.shtml

5. Demo of record editing

Check some record, e.g. Barbara's thesis. Login as admin. See Edit this record link. Click on field ordering. Click on text to enter it. Press m to go to selection mode, drag mouse, press m again, and press DEL to delete many fields at once. Click on Help to show hotkeys. Speak of editing efficiency. Press Cancel not to submit changes.

URLs in this demo:

• http://inspiredev.cern.ch/record/edit/

6. Demo of record merging

Take 0804.3349 (recID 764167) and its new clone record (recID 772788). Note title change, replace the field from rhs to lhs. Note abstract change, copy it from rhs to lhs. Speak about publisher feeds and the provenance tag ($9). Note author differences, show merge of all non-conflicting fields in one go. Press Cancel not to submit changes.

URLs in this demo:

• http://inspiredev.cern.ch/record/merge/#recid1=764167&recid2=772788

Demo No.3 (Apr 2010)

Demo given by Tibor at the 4th HEP Information Resource Summit, Cambridge, 2010-04-15. Showing just-released INSPIREBETA, as well as some forthcoming features (plots, full-text snippets).

See also slide deck attached below for slides shown at this summit.

Part One: From SPIRES to INSPIRE

1.1 Keeping users at home

Show SPIRES search interface. Show INSPIREBETA search interface. Important for users to feel at home. Copy SPIRES search example:

FIND T QUARK AND A RICHTER, BURTON AND NOT DATE < 1984

Show it works on INSPIREBETA. Talk about spending time reproducing SPIRES features for users to feel at home.

But we can already see e.g. one novelty of INSPIRE: Google style free keyword search. rowson heavy quark 2580 leading to the same result. Freely combined author, title, page number search.

1.2 RSS feeds

Another INSPIRE goodie example: RSS feeds.

Example one: researcher interested in neutrino oscillations topic. Search for neutrino mixing. Show bottom link ``Interested in being notified...?''. Subscribe to the RSS feed, show how things work with Firefox LiveClick. Alerts, mobile devices.

Example two: Pretend being author of a record from the above neutrino mixing query that has some citations displayed. E.g. arXiv:1003.5800 having two citations. Search for this arXiv number in the reference field. See there are two hits indeed. Show that RSS alert for this query will now behave quite like Elsevier's Cite Alert for this individual paper.

1.3 Cite summary

Another INSPIRE goodie is that cite summary is now much faster. So we can do stuff like search for hep in reportnumber and show complete cite summary.

So let's take Harvard in affiliation as an example. Follow top cited papers (Renowned category) link. Take 4th hit (``Unity of All Elementary Particle Forces'') as an example to show another INSPIRE novelty: the detailed record pages.

1.4 Detailed record and citations

So we reached http://inspirebeta.net/record/92111/. Speak of tabs. References, Citations. Browsing in the citation network. Co-cited with functionality, discovery of papers. Show citation history graph. Citation peak ~10 years after publication, then plateau. http://inspirebeta.net/record/92111/citations

Mention other citation graph patterns. E.g. sleeping beauties. Take one of the famous old papers that gets late citations, wigner thermodynamic equilibrium, paper from 1932. Show citation graph, confusing our x-axis manager. Sleeping beauty effect for ~40 years, now about 20 citations per year. http://inspirebeta.net/record/3198/citations

Note: nice citation graphs can be seen on the top two papers of Peter Minkowski. So one can start from the author page http://inspirebeta.net/author/Minkowski%2C%20Peter and click on top cited papers, first one having a camel-case shaped citation history graph http://inspirebeta.net/record/1774/citations and second one having an impressive sleeping beauty citation history graph http://inspirebeta.net/record/4994/citations.

1.5 Author page

Another example of INSPIRE novelties is the dedicated author page. Search for dixon in author index. Click on the 4th hit having full name Lance J. Dixon leading to the nice author page Lance J.. Show various summarizer (aka second order operator) result boxes. Keywords, co-authors. Affiliation history. 90% papers in HEP are theory, 2-4 authors, relatively easy to maintain affiliations. But what about big experimental collaboration papers?

1.6 CMS publication lists

Example of a recent big experimental paper: search for CMS-QCD-09-010. Detailed record browsing joy http://inspirebeta.net/record/845323. 1,968 authors from 172 institutes. How do we manage that?

Show XML format discussed with collaborations http://inspire-hep-dev.cern.ch/img/cms.xml. Explain authorIDs, persistent across institute change. Explain HEP names history, getting XML file from CMS, entering and verifying authors, enriching XML file with INSPIRE author IDs. Willingness of collaborations to incorporate INSPIRE author IDs back to their internal databases. Hint at exchanging information with publishers, hits at the ORCID initiative.

1.7 User feedback

These were glimpses of some novelties of INSPIREBETA on top of keeping existing SPIRES functionality and user experience. Soft released two weeks ago. Show glimpses of a typical user feedback: #81955. People generally like speed, and generally miss not-yet-fully-ported features, such as H-index.

2. Part Two: glimpses on the forthcoming features

INSPIRE built on top of Invenio. Invenio has many features (like personalization) that were purposely hidden in order to ease the migration from SPIRES. Features to be adapted and progressively released to the community as `presents'. Two glimpses of the very forthcoming presents: plots and full-text snippets. Switch visibly to the DEV box.

2.1 Plots

Search for neutralino in caption index http://inspire-hep-dev.cern.ch/search?ln=en&p=neutralino&f=caption. Show second CDF and D0 paper http://inspire-hep-dev.cern.ch/record/719460?ln=en. Show plot slider. Go back and click on the first plot to go to the beginning of the Plots tab http://inspire-hep-dev.cern.ch/record/719460/plots. (any plot link to be fixed soon). Show plots and LaTeX formulas.

Go back to search. Select `HTML detailed' view to see many records with many plots on the page. Go slowly down the page while images are loading.

Interesting technical issue: plots currently attached to papers. But we have another use case: e.g. ATLAS collaboration having plots as standalone objects, with prior approval workflow, then reused in many presentations later. So, do we have paper-centric or object-centric point of view?

2.2 Full-text snippets

Search for qcd monte carlo in fulltext http://inspire-hep-dev.cern.ch/search?ln=en&p=qcd+monte+carlo&f=fulltext. Scroll down rapidly in order to `hide' the first published hit (that will come in the snippet demo finish). Show arXiv full-text snippets, speak about plot extraction.

Scroll down to bottom to reach CERN-THESIS-99-050. Speak about pre-arXiv era, willingness to self-archive from CERN theses author mailing project. INSPIRE as a place to put pre-arXiv material.

Scroll up to the first hit. Lecture Notes in Physics, published by Springer. Collaboration/partnership between INSPIRE and Springer, showing snippets courtesy of Springer.

So having snippets from arXiv, pre-arXiv, publishers. Mutual benefit for all actors.

2.3 Finale: combined search capabilities

Last example for the demo finale: the power of combining all the goodies together. Search for author:ellis caption:model cited:10->20 reference:astro. So we have combined metadata, figure caption, citation count and reference section search. Show second hit, astro-ph/0604272. Hint at ADS and INSPIRE common interests and curation needs.

Demo No. 4 (Dec 2010)

The overriding story here is that we want to have a lot of ways to slice the citation graph, to help us explore the literature and to understand an individual's body of work. Focus on smooth, tic-free delivery.

SPIRES: find a silverstein, e

INSPIRE: find a silverstein, e

    same results
    detailed record page: abstract, references, citations, plots!
    plots bug in LaTeX parsing
    PDF from arXiv - easy to see how people get confused

INSPIRE: find a silverstein, e

select citesummary
	very well known papers; click
	uncheck xCited; select author page - shamit kachru co-authorship.  note conflict w/ Ed Silverstein

refersto: silverstein

	RSS feed!

Eva works at SLAC. Let’s see what kind of place SLAC is.

• What have people been citing slac for lately?
  • refersto:aff:slac
• ...and what’s the most cited thing referring to SLAC?
  • check "times Cited"
• ...so what’s the most cited thing, cited by SLAC?
  • citedby:aff:slac
• ...and what have SLACers been citing recently?
  • uncheck "times Cited"

Close academic bonds may be gleaned from reciprocal citing behavior - like friending on social networks. Let’s see who at Fermilab is academically close to SLAC:

• refersto:aff:slac and citedby:aff:slac and aff:fermilab

clearly this is a simplistic example, but it gives us some insight into the citation graph.

Last example. (I think this is pretty cool.) CMS has different data streams. One is called highPurity - but not in the metadata.

• Search highPurity; you won’t find it.
• fulltext:highPurity

	first paper: note fulltext is from JHEP
	detailed record page
	files tab - math tables
	We can obviously link to just about anything here, and host almost anything too.  mathematica, root, maple, code.  this paper’s additional table is easy, but hints at the potential.