Wrap-up: Functionalities: MS Search Slider Bars

[shariboodts]

Based on example: https://passim.rich.ru.nl/manuscript/list?usersearch=9818 (Manuscripts > Collection/Dataset… > Historical Collection > Wolfenbüttel (irisdenis))

Part one of the issue:

• It is unclear to us how the slider bars work exactly? Menna’s old notes contradict what we see when we use them. I assume that they represent a percentage of the Historical Collection we want to compare with the manuscripts (i.e. “give all manuscripts that have at least 75% overlap with the content (AF’s) of Historical Collection ‘Wolfenbüttel’”). The results are based on the number of AF’s linked to the Historical Collection and to the manuscripts in question (similar to the DCT). Is this correct and does it apply to the different slider bars?
• Currently, only the bottom slider bar (Authority_File filters: HC/Manu overlap) seems to work, but it is unclear how it functions exactly:
• The bottom slider bar gives confusing results. 100 per cent overlap with the HC ‘Wolfenbüttel’ should only be Wolfenbüttel, Weiss. 12. Example: an overlap of 95 % still gives 13 manuscripts, including Weiss. 12 but also other manuscripts that have far less sermons (e.g. Firenze, Conventi Soppressi 137, that has 28). How can a manuscript with 28 sermons be more than 90 per cent equal to a much larger collection of 97?
  • It seems that the percentage of overlap is calculated from the point of view of the manuscripts: a manuscript with only 3 AF’s, which happen to all be also present in HC Wolfenbüttel is given as a result of the search for at least 95% overlap.
  • This is only part of what we want. It should be possible also to calculate overlap the other way around: In that case, we want the calculation to start from the content of the HC: only those manuscripts that have 95% of the AF’s present in HC Wolfenbüttel should show up in the result view.
  • The user should be able to specify: 95% of the manuscript content vis-à-vis the HC OR 95% of the HC content vis-à-vis the manuscripts.
• What are the other slider bars supposed to do?

Part two of the issue:

• The current system of choices, sliders and filters is confusing and there seem to be redundancies. What we actually want:
  • Three types of comparison should be possible via slider bars:
    • A: Compare AF-content of HC with manuscripts (a) and vice versa (b, see part 1)
    • B: Compare AF-content of one manuscript with other manuscripts
    • C: Compare AF-content of Personal Dataset (AF) with manuscripts
  • User makes a choice: one of these three comparison tools, when clicking on it, one slider bar becomes visible. User then makes a second choice: percentage applies to the selected item (HC, specific manuscript, PD (AF)) or percentage applies to the results (see part one) (can be via a button to click?).

NOTE: If this requires too much work, then we choose just one option > percentage will then only apply to the source item (HC, specific manuscript, PD (AF)).

So (overview):

o Manuscript comparison keep o Historical Collection keep (same as HC/Manu overlap? We need clarification!) o PD: Manuscript discard o PD: Sermon manifestation discard o PD: Authority file keep o HC/Manu overlap can be discarded (is this one indeed redundant?)

Part three of the issue:

• The possibility to select manuscripts to compare should be simplified. o Collection/Dataset... as field should be renamed: Comparative search o Manuscript comparison > “Select multiple manuscript identifiers...” only allows to search for manuscript identifiers (i.e. shelfmarks), without cities or libraries. Users should be able to get a drop-down list of manuscripts (city-library-identifier) when entering only the city (“Orléans”) or library (“badia”), in addition to a simple search by entering an identifier/shelfmark. It should work like the option to add manuscripts in a DCT (Source > Manuscript > “Select manuscript...”).

• Change description: ‘Value: (move slider)’ should become “Percentage of overlap (move slider)". Number should be visible while sliding so you can make an accurate choice (now, empirically you see the percentage only when you stop moving the slider)

[ErwinKomen]

Looking into part 1

1. Question: what does the slidebar do?
   1. Location: ManuscriptListView
   2. Note: there is only one slider defined
      1. The defined slider's info should be made visible. It says: Percentage overlap between the Historical Collection SSGs and the SSGs referred to in the manuscripts
      2. The fact that four sliders are visible is a bug

Implementation notes

1. Fixing visibility from occurring in four places to occuring in just one place, i.e. the part of the head_id, which is filter_collection:
   1. Corrected basic_list, which shows hidden (head_id) just anywhere (line 269, 'show the different filters in the list')
   2. Removed a superfluous hidden head_id line in ManuscriptListView
   3. This resulted in the slide bar only occurring inside the Collection part
2. The visibility of the sliding bar is as follows:
   1. It is shown by itself, if HC/Manu overlap is selected
   2. it is automatically included, if any of the following are selected: Manuscript comparison, Historical collection, PD Authority file
   3. It is not shown automatically when these are selected: PD Manuscript, PD Sermon
3. Location of the slider in the listview: it is filter collection_hcptc
   1. It makes use of dbfield hcptc, which is a non-existing field
   2. The actual overlap is calculated in the view's method adapt_search(), looking for overlap and making use of get_overlap_ptc()
   3. This shows that via field manu_colloverlaps the model CollOverlap is used.
      1. That model contains combinations of: 'collection', 'manuscript', 'overlap'
      2. The percentage taken there is determined by taking the HC as a starting point, and counting the number of SSGs matching an item from the HC.
      3. However, I need to check whether the set of SSGs that are 'part of' a manuscript is calculated correctly
         1. Yes, get_overlap() correctly calculates this, taking the HC as a starting point
      4. But when is get_overlap() actually called?
         1. Well, it is called whenever a comparison (with slider) is being asked for.
   4. Okay, I double checked, and the table CollOverlap is filled completely correctly.
   5. But the culprit is this: the calculation of the Q-expression in fields['overlap'] in method adapt_search() in the Manuscript list view. That did not include the list of collections with which a comparison is made, nor the user's profile.

Comment on part 1

The wish of the user was:

The user should be able to specify: 95% of the manuscript content vis-à-vis the HC OR 95% of the HC content vis-à-vis the manuscripts.

Well, there is one slider, and it has just one function in relation to the HC/Manu overlap when a HC is selected. And that one function is just one of the two above: the percentage of the manuscript content w.r.t. the HC. E.g: 70% of the HC's authority files occur in each of the manuscripts in the output. For Wolfenbuttel that would yield just 3 results (manu 1779, 7788 and 7217)

[ErwinKomen]

Part 2 - what is already there

1. Three types of comparison are already possible via slider bars:
   1. Compare AF-content of HC with manuscripts (a) and vice versa (b, see part 1)
   2. Compare AF-content of one manuscript with other manuscripts
   3. Compare AF-content of Personal Dataset (AF) with manuscripts

[ErwinKomen]

Part 2 - cleanup

1. I removed the HC/Manu overlap button, since it only shows the slider
   1. Implementation: add "hide_button": True in table filters and process this in basic_list.html
2. Discarded:
   1. PD: Manuscript discard
   2. PD: Sermon manifestation discard

Final checking

1. When "Comparative search..." is clicked, the slide bar must also always disappear.
   1. This is now not happening in the following situation:
      1. Click "Comparative search"
      2. Perform a Historical collection or Manuscript comparison using a particular percentage
      3. Click "Comparative search" again.
   2. Resolution: hide the class of range input items in filter_click()

[ErwinKomen]

Part 3 - other

1. Made range slider show percentage on moving
   1. See basic.js, trick was 'input' event
2. Changed slider value description
3. Possibility to select manuscripts: made it work as in the DCT example

[ErwinKomen]

Leftover

1. Not implemented is the option to calculate the percentage the other way around.