madewild / camille Goto Github PK

Use universal calendar??

An idea to be discussed later: we could add a "(Un)Select all" button, which could allow to go faster if the user only wants to (un)select one or two newspaper(s) (the same could be done for other filters with checkboxes). But this "(Un)Select all" button should work without activating the filter, otherwise it could be problematic.

Originally posted by @anjacque in #36 (comment)

track_total_hits=true for count api

Parse id and reindex ES

Like newseye, needed?

Add journal name in natural language etc

Using spaCy?

Cron job?

Add language field in ES to anticipate for NL docs

Quotes should be escaped in return form

Cf. IDLab demonstrator https://tw06v072.ugent.be/kbr/ and https://github.com/filipradenovic/cnnimageretrieval-pytorch

So that it will be by default the first file in the list, which is potentially 500 files long

Waiting for domain name...

Add basic stats in the zip (number of files, breakdown by journal, etc.)

With some explanations about the project

Allow to perform case-sensitive searches, e.g. only "jambes" without including the city of "Jambes"

On newspaper, year (cursor)...

It seems to work properly! Two things if we want to improve it even more:

1. We could make impossible to enter "impossible requests" in combination with other filters (for example, when the "Exact date" filter does not fit the years selected in the "Year Filter"). For now, it only shows "Aucun résultat". Could the different filters work in a dynamic way?
2. The drop-down menu for the year selection is not really practical to use because we have a large range of years. I don't know what to propose to fix it, except if we have a box to fill by entering a year manually. To be discussed.

Originally posted by @anjacque in #30 (comment)

Parse id and reindex ES

To allow larger page numbers (e.g. 500) which might be inconvenient in a slider.

In JB838 and JB1051

In addition to modal

Parse id and reindex ES

Explain Lucene syntax, results...

Possible with route53?

@anjacque can you provide them?

Parse id and reindex ES

To 60GB/node

For overview of hits across years

and booleans

input field with calendar (from - to)

The font for the text containing the selected years in the "Year" filter is not the same as the one for the rest of the page and the other filters.

Currently only possible one by one (newseye style), to replace by checkboxes or multiselect?

Add README file (+ basic LICENSE?) with some explanations about the contents (limit of 500 files, format, etc.)

Tweak css for font size/style

To save money once all XML are there. Hoping it will not crash...

Allow to search only with filters with no keyword (only pages of some journal/year/date etc)

Use * for that ?

CAMille will provide users with a set of external online resources related to the history of journalism in Belgium.

For the moment, only write "Under construction"

Exporting XML or TXT ? Limit ?

To show more than 10 hits

Sort by date asc/desc automatically on select

Elasticsearch managed service too expersive with 3 nodes

Define distance between two terms with AND

Allow to override default ranking (by relevance) to rank by date for instance

<S interpreted as strikethrough in https://dev.camille-ulb-kbr.be/?query=*&paper=JB421&sortcrit=relevance&year_from=1930&year_to=1930