Capitalize titles when all uppercase

[jmartinm]

When titles are all uppercase they should be capitalized (and only in that case)

Example

There is already a filter but it is not being used at the moment. https://github.com/inspirehep/inspire-next/blob/master/inspirehep/ext/jinja_filters/record.py#L454

[jmartinm]

Looking closely at what happens in current production, if a title is all in capital letters in the metadata, like:

RACETRACK: A COMPUTER CODE FOR THE SIMULATION OF NONLINEAR PARTICLE MOTION IN ACCELERATORS

On display time, all words get capitalized:

Racetrack: A Computer Code For The Simulation Of Nonlinear Particle Motion In Accelerators

Do we still want that, or we capitalize the sentence?

Racetrack: a computer code for the simulation of nonlinear particle motion in accelerators

cc @inspirehep/inspire-dir

[kaplun]

Wild idea: what if we build a statistical dictionary based approach (based on the correct existing title) and automatically correct these titles? E.g. we could:

• consider all the non-capitalized titles in INSPIRE
• ignoring the first word of sentences, build a dictionary with every word (counting how many times these appear).
• We can then query the dictionary and take the case of the most common word that matched.
• If a word don't match we can simply decide to make it lower
• Of course if it's at the beginning of a sentence we respect the case.

This could be a nice quick project to correct in general the case of all the old records. On average it cannot be worse than the current all-capitalized titles.

[annetteholtkamp]

Great idea

• Annette

On 09 Feb 2016, at 10:23, Samuele Kaplun notifications@github.com<mailto:notifications@github.com> wrote:

Wild idea: what if we build a statistical dictionary based approach (based on the correct existing title) and automatically correct these titles? E.g. we could:

• consider all the non-capitalized titles in INSPIRE
• ignoring the first word of sentences, build a dictionary with every word (counting how many times these appear).
• We can then query the dictionary and take the case of the most common word that matched.
• If a word don't match we can simply decide to make it lower
• Of course if it's at the beginning of a sentence we respect the case.

This could be a nice quick project to correct in general the case of all the old records. On average it cannot be worse than the current all-capitalized titles.

— Reply to this email directly or view it on GitHubhttps://github.com/inspirehep/inspire-next/issues/847#issuecomment-181775516.

[ksachs]

Bad examples:

Diffraction And Precise Qcd Measurements At Hera [https://inspirehep.net/record/1422853?ln=en]

Soft Qcd At The Lhc [https://inspirehep.net/record/1422855?ln=en]

I think it's the hole conference: [https://inspirehep.net/search?ln=en&p=773__w%3AC12-03-10&ot=245&action_search=Search]

[ksachs]

Looking at the Moriond contribution titles, could we keep it simple?

High Order Qcd Predictions Of The Higgs Boson Production Cross-Section At The Lhc

is almost unreadable. I think non-capitalized titles are better

High order qcd predictions of the higgs boson production cross-section at the lhc

With a KB of (basic) acronyms and the names of collaborations from the EXP collection ->

High order QCD predictions of the higgs boson production cross-section at the LHC