# June 2019 tag clean-up initiative

On June 22, (anticipated lowest activity on the website), the required edits\retagging bumped a lot of questions to the main page. Last time it happened a couple of years ago when this activity was held. Keeping track of the questions, tags, newly created tags in the real-time would help the community to avoid such "clean-up"s. To view for active and recently asked questions, use also Hot Today and Hot Weekly tabs on the main page.

Computations Science SE Community has not done a tag clean-up for a while.
Tagging helps to organize the questions and keep track of the interesting ones. With more than 7500 questions, Computational Science SE certainly is in the phase when tagging is important.

As per this request, we probably should do an annual tag clean-up initiative. Let's see how it goes this time.

Goals:

• identify misspelled tags or tags that differ by the plurality (ex. matrix/matrices), word order, abbreviation, etc.
• merge AND synonymize tags with equivalent meaning.
• synonymize (no merge) tags with the close scope and relatively low usage. This would allow to separate them in the future if needed.
• identify the tags that do not help in question categorization

Method:

• 1 proposal per answer is recommended to simplify the discussion. I will group some of my suggestions.
• An ideal proposal contains the desired action (delete, synonym, merge, describe) and a brief motivation.
• Similar to the previous initiative, let's collect the suggestions for around a week (till June 15) and proceed with actions on June 22 (giving another week for revisions/voting).

The more people vote/suggest, the easier it would be for the moderators' team to not do mistakes and organize things better. I personally use tags a lot and would benefit myself greatly from better content categorization.

• I wonder if we can get a hand from devs to get a CSV file with tags and use to do some analytics on it. – nicoguaro Jun 5 '19 at 0:11
• @GertVdE was helping the last time with the tag clean-up. I wonder if he was using something custom-made. – Anton Menshov Jun 5 '19 at 0:13
• @nicoguaro this query might help: data.stackexchange.com/scicomp/query/1060967/… and would allow export to CSV – Anton Menshov Jun 5 '19 at 2:02
• Great news! I will take a look tomorrow. – nicoguaro Jun 5 '19 at 2:07
• Last time, I worked on a CSV file provided to me by either Geoff or Aron, – GertVdE Jun 6 '19 at 23:14
• @GertVdE I think my query (in the comment above) will produce exactly the one that is useful. – Anton Menshov Jun 6 '19 at 23:15
• @AntonMenshov:yes, perfect. but somehow I was too noob then or didn't have enough reputation to get to that query myself ;-) – GertVdE Jun 6 '19 at 23:17
• As a general remark: once we have pruned/cleaned the tag list, there are two final actions to be done, I think. First one is go through all final tags and see if the description is adequate. Second, we should see if questions need to be retagged (that might be too much of work, could be limited to questions with score > 10 or > 5). As a final call: we (moderators, long-time users, all) should pay more atttention to tagging when a new question is posted on SciComp. – GertVdE Jun 7 '19 at 9:53

Company-wide tag . Ansys has multiple products that are relevant to the computational science world. At list has its own tag (which is a separate topic).

Currently, we have 3 questions with this tag.

Suggestion:

• delete from all question without creating any synonyms/merges due to aforementioned reasons.

Reasoning:

Simply merge tags with identical or close meaning:

Delete rarely used tags with poor potential to get more attention in the future:

which currently clutter the tag space.

A group of biology-related tags. In our community, we are certainly focused on which is a relatively popular tag (25 questions as of June 4, 2019).

Suggested actions:

After that, we will have a relatively healthy 30 questions in the , most of which will still feature at least one other tag.

Our old friends: and , both used 5 times in 9 questions total.

• Or regarding mesh elements, where a much more appropriate tag is already used:
• or regarding the Boundary Element Method

Question 8

The last question uses tag certainly inappropriately anyway.

Suggested action:

With all that in mind, I do not think that we should use any merging/synonyms here.

• Those tags are, IMHO, misused now and do not help to organize content. So, they should be deleted from the aforementioned questions with appropriate manual retagging where needed.
• I need to educate myself on SE tag policy because we might use some action to prevent the usage of those tags. The easiest would be to add a proper description tags and with bold "do not use, please select the more specific tags, for example ,..". Not sure if we can do more than that.

This is the distribution of tags

with

• Median: 19.5
• Interquartile range: 55.25
• Mode: 3

The following wordcloud shows the relative frequency of tags.

What is the added value of the tag? It is a highly used tag (476 questions as of June 7th, 2019) but always in the context of linear-algebra. However, only 218 of 476 questions have the tag . Out of 476, 258 questions are tagged with but not with .

It seems that the tag is also used as a generic tag like . The questions have different topics: sparse matrix handling and multiplication, solving linear systems, matrix functions, ...

Suggestions

1. retag questions tagged with with and remove the tag .
2. after that, review questions that have as only tag and add a more specific tag appropriate for the question.

In linear algebra, one considers dense and sparse matrices. For the former, the tag is , for the latter .

Suggestion: rename to

• I started to look at sparse and the description right away is talking taht this tag is not limited to sparse matrices with a (unfortunately rare for our community) tag-wiki describing what is implied. What are your thoughts on that? Separate sparse-matrices from everything else? – Anton Menshov Jul 5 '19 at 1:56
• From a (very) quick browse through of the questions tagged sparse: they all are about sparse matrices. I don't see any question (again, a quick browse) on a subject on sparsity that is not matrix related... So I would rename the tag and modify the tag description accordingly. – GertVdE Jul 5 '19 at 6:26

The tag (51 usages) is used for multiple topics covered by other tags:

Suggestion: not yet, open for discussion.

• This is a huge one. But I am totally for removing random completely and retagging the questions to the appropriate categories. My reasoning: the more precise tags are pretty widely used (so we are not splitting hairs here) - and describe the question much better. I, in particular, will not be interested at all in questions on stochastic-ode, but others might be of my genuine interest. – Anton Menshov Jun 7 '19 at 22:25
• Do we have questions related with random-number-generator? – nicoguaro Jun 12 '19 at 15:59
• @nicoguaro: yes, multipe ones: Q1,Q2, Q3, ... just search for "random" and "period", for example – GertVdE Jun 12 '19 at 18:29
• @GertVdE I did some automatic and manual retagging of the questions. There are 11 questions left tagged random. Can you help retagging those manually in the appropriate categories? – Anton Menshov Jun 22 '19 at 22:53

Rename the tag to . The bulk of SciComp is about algorithms, no need to add that in a tag.

There exists a tag and a tag . All questions with the tag are about matrix factorizations. I don't see that advantage of having two tags (even as synonyms)

Suggestion: retag all questions tagged with and remove the tag .

The tag is used both for quadrature/cubature and for solving ODEs.

Suggestion: re-tag all questions that have with either

In the tag description disencourage the use of the tag

• A bit controversial here. Not all integrations that are relevant to Computational Science are quadrature-based. For example, I would tag this question with integration and would not tag it with quadrature. Thoughts? – Anton Menshov Jun 7 '19 at 22:31
• @AntonMenshov: which question? I think you forgot to hyperlink it... – GertVdE Jun 10 '19 at 20:24
• fixed :) This whole post made me learn all the shortcuts for links, references, and tags. – Anton Menshov Jun 10 '19 at 23:55
• I understand. Quadrature, to me, is synonym for "numerical integration". We could allow for the "integration" tag to identify questions on symbolic integration or "problems associated to integrals". But as soon as there is a numerical evaluation component, I would tag it "quadrature". Or if quadrature has too much a 1D connotation, create a tag "numerical-integration" and have "quadrature" as a synonym. I would definitely split off "differential-equations" from "integration". – GertVdE Jun 11 '19 at 8:10

A general for the main site tag (5 usages as of June 4, 2019).

Here, as per Meta post on Career advice, one question will be retagged with , and for 4 others, this tag is proposed to be simply removed.

I will also add to my bucket list an item to search for posts relevant to , as certainly, the aforementioned one is not alone in this category.

• That tag looks so bad that I would suggest blacklisting it, not simply deleting it. – Federico Poloni Jul 2 '19 at 18:13

The has 4 usages, out of which 2 are closed.

Usually, the questions with this tag would be a poor fit for our community, and I do not see why a particular tag would be useful.

Suggestion:

• Leave tag alone as it is.
• Add a description discouraging its blind usage and suggesting to look at the famous on-topic section before asking.
• I would remove it and label the questions using it with software. – GertVdE Jun 7 '19 at 7:04

Searching on the word data gives the following tags (ordered in decreasing popularity):

My impressions follow...

### data-visualization

Clear. Questionso on how to generate nice plots of data.

### data-analysis

Clear. Questions on algorithms to manipulate and interrogate data.

### data-sets

I see at least two different interpretations:

1. Questions on where to find items (I try to avoid the word data here) to test algorithms. These items can be a set of test functions (like the Rosenbrock function in optimization, like the IVP test set of CWI, like the Gautschi test set for quadrature, ...)
2. Questions on where to find items what are in my opinion data sets to test statistics, machine learning, classification, ... algorithms (like sensus data, like world population data, ... )

Suggestion 1: split this tag in and

### data-management

Multitude of interpretations:

1. Data storage
2. Pipelines or methodologies to handle/interpret large data sets
3. Software specific questions on HDF5, Pandas, PyTables or requests for recommendations

Suggestion 2: detailed analysis of the questions tagged with and possible retag.

### data-structures

Clear. Questions on optimal data-structures for algorithmic applications.

### data-storage

Again a mixture of different topics:

1. Hierarchical data formats
2. Data storage
3. Use of data structures in algorithms

Suggestion 3: detailed analysis of the questions tagged with and possible retag.

### spatial-data

Clear, but low usage (only 9 questions).

• any thought on data-visualization and visualization? – Anton Menshov Jun 7 '19 at 15:21
• I consider that visualization is the right tag. I suggest to retag and remove data-visualization. – nicoguaro Jun 7 '19 at 21:35
• @nicoguaro agree. "data" in "data-visualization" seems redundant. Anything that is visualized is data. Actually, most of the things are data. – Anton Menshov Jun 7 '19 at 22:15

The tag (11 questions tagged) is being used for totally different topics:

1. Assembly of matrices (9 questions)
2. The assembly low-level programming language (2 questions, Question 1, Question 2

Suggestion: remove the tag from the questions tagged with and reserve the tag for the low-level assembly programming language.

• Yep, this one will certainly need a proper description due to ambiguity. – Anton Menshov Jun 7 '19 at 14:56

The tag seems a bit superfluous. It is used in 18 questions and, according to me, they are all on the topic of .

Suggestion: retag all questions tagged with and remove the tag .

# Tags related to text

### documentation

Clear. Questions on how to write documentation. Although Geoff posted some questions on reproducibility where he also included this tag. I kind of disagree to use the tag for these questions.

### publications

Clear. Questions on good practices to publish results.

### journals

Clear (in my opnion): questions on which journal might be more appropriate for a certain research paper or journals to read to follow a particular research domain.

### reproducibility

Clear. Questions on workflows, pipelines but also good practices on how to write reproducible papers.

Suggestion: reserve the tag for questions on code documentation (good practices).

• I would merge journals and publications. They are often used together. Do you see the need to separate the two? Otherwise, I would totally agree. – Anton Menshov Jun 7 '19 at 14:59
• OK. I would opt for "publications" – GertVdE Jun 10 '19 at 20:29
• Merge performed. Now we are missing tag info\guidance for publications, journals, and reproducibility. In the to-do list, but if anybody takes an initiative, it would be only welcomed. – Anton Menshov Jun 22 '19 at 19:38

Cluster-computing is in my opinion umbrellad by high-performance-computing. I'm not a big fan of abbreviations but I think HPC has become more a word than an abbreviation.

Suggestions

• there is also cloud-computing. Any thoughts on that one in that regard? Synonym for now? or you would consider it to be a separate enough entity? – Anton Menshov Jun 7 '19 at 14:55
• @AntonMenshov, for now, I would put everything under hpc. If more questions are posted specifically on cloud-computing, the tag will re-appear automagically :-) – GertVdE Jun 10 '19 at 20:33

# Precision vs accuracy vs floating point/fixed precision

According to my dictionary:

Accuracy = nearness of a calculation to a true value.

Precision = resolution of a result (how many correct digits)

Relevant tags:

No tag for arbitrary precision (although some questions explicitly mention this, for example Q1, Q2).

The tag is used for both concepts, but also in the context of floating point precision (IEEE 754 single/double), fixed precision (implementations like Gnu MultiPrecision library).

The tag has to my opinion little added value. Most of the questions tagged with this also tag with or .

The tag (9 questions as of June 7, 2019) is linked to the above concepts.

Suggestion: I don't have a clear suggestion as of today. Only created this to get a discussion on-going.

• That needs to be addressed. I also don't have a clear plan of action. – Anton Menshov Jun 7 '19 at 22:21

In many cases and are interchangeable. I do sense a semantic difference. To my feeling, efficiency seems to be "lower level", i.e. the efficiency of a fundamental calculation (a BLAS DGEMM, for example) while performance seems to be of a whole algorithm or simulation (often related to HPC).

The tag could also be linked to these two.

Suggestion: make a clear distinction in the tag descriptions (even hinting which tag should be used in what context) and review all questions tagged with either or and re-tag if deemed necessary.

I believe this activity is hugely ill-conceived! You have completely polluted the "Top Questions" view of the main page with your minor edits. I can't tell where the first "real" question is amongst all this "cleanup". As far as I'm concerned, you are not adding value with these edits but rather just the opposite.

• surely understand the frustration. That's why it happened on Saturday in the summer, the lowest-activity time. Note, that for a long time, there was little work done on the tagging. Ideally, the organization should happen "not after", but that's the reality we live in. – Anton Menshov Jun 22 '19 at 23:11
• There are also Hot Today and Hot Week tabs on the main page, where you can easily find the recent-activity questions - which would filter out all th activity you are not interested in. – Anton Menshov Jun 23 '19 at 1:03

This answer serves as a report for the initiative and a list of remaining actions.

Note, this initiative has a drawback of the over-activity on the main page of Computational Science SE, so it's better to have it during the weekend when the activity on the site is minimal.

Not yet done:

Done:

What do we do with and . Again, to me there is a semantic difference. But I see that a reasonable number of questions that have been tagged I would have tagged .

Suggestion: Not yet. Open for discussion, I would say...

• I would suggest using visualization. plotting would make sense in StackOverflow, when asking how to do a specific plot. – nicoguaro Jun 7 '19 at 21:39
• @GertVdE, can you formulate a semantic difference that you see? – Anton Menshov Jun 7 '19 at 22:20
• @AntonMenshov, for me plotting is about the technical part ("how to plot ... with matplotlib", for example), data visualization is more conceptual ("what type of graph would be best to display this type of data?"). – GertVdE Jun 10 '19 at 20:32

Some questions clearly indicate that it is a homework problem and the OP is asking for hints to get started or verification of her solution. Other questions explicitly state that it is not a homework problem. I do appreciate knowing beforehand that a question is related to a homework problem because I my answer would be tailored to this information.

Suggestion:

• Introduce a tag and tag those questions that explicitly state that is home work.
• In the tag description encourage the use of the tag
• Remove the phrase "This is not homework" by editing the relevant questions.
• scicomp.meta.stackexchange.com/a/325/20688 "the death of meta tags" – Anton Menshov Jun 7 '19 at 15:18
• this tag previously existed on Computational Science SE and has been eliminated. – Anton Menshov Jun 7 '19 at 15:19
• Ok. If that is the agreement on SE. I beg to differ:-) – GertVdE Jun 7 '19 at 15:36
• It probably would be a separate discussion later. And we might want to look at what Physics, Chemistry and Math do about their homework-type questions, as well as their discussion on the "meta-tags". – Anton Menshov Jun 7 '19 at 15:51
• Agreed. Good plan. – GertVdE Jun 7 '19 at 16:54