softvare-group / mibtex Goto Github PK

It seems that no citations are found anymore.

Entries such as SGB05 and KG06 are simply not found as google scholar omits there subtitles. If a paper cannot be found, it should be rather easy to just check the title again, but removing everything starting from ":".

As the retrieval of citations heavily depends on the keys, it would be great to have a check when parsing the file for duplicated keys. In such a case, no export should happen.

https://scholar.google.de/scholar?q=Partially%20Safe%20Evolution%20of%20Software%20Product%20Lines

Is found, but has no citations so far. As result, I would expect 0 as the citation count.

As in the multiple HTML version.

Maybe in the top right corner. Something like "12/781".

We should improve the parameter handling through a config file or flags in the command line call.
This makes it even easier with optional parameters.

Would then be a similar behavior as in the multiple HTML version.

If something goes wrong while updating the citations, -1 and the current time stamp is stored. Instead, it should rather be untouched.

Proposal for citation numbers:
0 or larger means that the entry was found and indicate the number of citations
-1 means that it was never checked (initial value of every new entry)
-2 means that it could not be found on Google Scholar
-3 means that something went wrong when checking the entry (could indicate a bug in MibTeX or a problem with that particular paper title)

Ideally, whenever exporting to HTML_NEW the citations.csv would be updated accordingly. However, no already computed values should be removed. No longer existing keys should be removed and new ones be added.

Calculation: Citations / (Current Year - Year of Publication)

If "Current Year - Year of Publication" is zero or negative, then keep the citations as-is

If Citations are negative (not available), keep them as-is

I would propose to use one digit more after "." (e.g., 10.3 for 31 / 3)

The process of adding new BibTeX entries as well as downloading and renaming papers involves numerous manual tasks:

1. Searching for a BibTeX entry in the web in a particular order (ACM, DBLP, Springer, ...).
2. Cleaning up the BibTeX entry (see example below for ACM)
3. Allow manual modifications to the entry, such as adding tags
4. Inserting the new BibTeX entry into an existing BibTeX file
5. Searching for a PDF and downloading it to a particular folder with a certain name

@inproceedings{Fischer:2016:PEA:2897010.2897011,
 author = {Fischer, Stefan and Lopez-Herrejon, Roberto E. and Ramler, Rudolf and Egyed, Alexander},
 title = {A Preliminary Empirical Assessment of Similarity for Combinatorial Interaction Testing of Software Product Lines},
 booktitle = {Proceedings of the 9th International Workshop on Search-Based Software Testing},
 series = {SBST '16},
 year = {2016},
 isbn = {978-1-4503-4166-0},
 location = {Austin, Texas},
 pages = {15--18},
 numpages = {4},
 url = {http://doi.acm.org/10.1145/2897010.2897011},
 doi = {10.1145/2897010.2897011},
 acmid = {2897011},
 publisher = {ACM},
 address = {New York, NY, USA},
}

@inproceedings{FLRE:SBST16,
	author = {Fischer, Stefan and Lopez-Herrejon, Roberto E. and Ramler, Rudolf and Egyed, Alexander},
	title = {{A Preliminary Empirical Assessment of Similarity for Combinatorial Interaction Testing of Software Product Lines}},
	booktitle = SBST,
	isbn = {978-1-4503-4166-0},
	location = {Austin, Texas},
	pages = {15--18},
	doi = {10.1145/2897010.2897011},
	acmid = {2897011},
	publisher = ACM,
	address = NY,
	year = 2016
} 

@String{SBST = "Proc.\ Int'l Workshop on Search-Based Software Testing (SBST)"}

It would be nice to have a tiny GUI, which automates these tasks as far as possible.

Currently, it takes some time to type something into the filter box. Ideally, all letters would appear immediately and the current filtering is aborted.

Worked already in the single HTML version.

Sometimes, the URL of an entry is just the doi lookup. It would be useful to add an option to filter such URLs in the cleaner (similar to how we can currently filter entire fields currently). Such a filter would be useful for submissions to ACM or DFG which need DOIs but we could save space when omitting duplicate URLs.

Currently, we can only give as parameter which BibTeX taglist (e.g., tt-tags or rs-tags) is used for the row with Tags in HTML. It would be nice to use multiple BibTeX taglists for the output.

If multiple taglists are choosen, starting from the second one the name of the taglist should be used as prefix in the following form:

tt-tags = {Bla},
tc-tags = {Blub},

"Bla, tc-Blub" if tt-tags is the first taglist in settings file
"Blub, tt-Bla" if tc-tags is the first taglist in settings file

When citation service runs in background, it may happen that citations.csv is updated, but these changes seem to be overridden. Also, can we make sure that nothing bad happens if both processes want to write the file at the same time?

Citation -2 instead for actual citation for many entries, which have been correctly found before, including:

DDP:EMSE17
DABW:STTT17
CBA:OOPSLA17
DL:iFM16
HNTS:ISoLA16
MTS+:ICSE16
...

Entries still not found:

MKR+:ICSE16

Added a filter and pressed enter. Filter works as expected. However, after a refresh the keyword is still included (which is intended), but all results are shown.

I would recomment to use the edit distance and check that it is smaller then titleLength/5. Otherwise, store -2 as the paper is essentially not found, then.

Currently, the order of citation entries in citations.csv seem to be reordered all the time. That makes it hard to compare changes via diff and to merge new numbers retrieved by different machines.

For example, they could always be ordered based on the Bibtex keys.

I corrected a wrong title and the citations file was not updated at all. Hence, the citation service is still querying google scholar with the wrong title.

It would be great to have a button/link to the source file (literature.bib) on the top of the HTML page.

Also, add the filter boxes to the table and let column width be decided by the browser.

Clicking on the paper titel should open the PDF. Clicking on an author name should start a google search with the author name (just the single name, not all authors of the paper). Clicking on the number of citations should search for the paper titel in Google scholar (as you do for retrieving the citations).

Currently, -1 is the result if a paper was found, but not yet cited (e.g., BL:FOSD14).

Currently, the sorting does not seem to work.

First of all, the results are very good now! I actually found almost only wrong entries in Google Scholar and only a minor issue with the following example:

https://scholar.google.de/scholar?q=Variant%20Comparison%20-%20A%20Technique%20for%20Visualizing%20Software%20Variants

Maybe it would be good to compare titles after removing all white-spaces or even other signs that are not letters. What do you think?

The issue has time. Fix it somewhen when there are no urgent thinks to do for FeatureIDE.