Hi ,
This is an idea that occurred to me while reading the stories. (based on anecdotes)
While reading
https://storyweaver.org.in/stories/131-babachya-mishya
I found that Marathi translation has more tags and many of them very good, but Hindi translation has only 3 tags.
So first step For stories with multiple translations, merge tags (english) from all translations .. These will form the candidate set of tags
Next we can take page wise content and can use wordNet or conceptNet to find a confidence score of these tags to sentences/page.
Top scoring tags can be suggested to end reader for validation while they are reading these stories on website. They can validate a tag by simple up/down method. And this can further validate.
So overall approach
Seeding the tag from all translations
Assigning scores
book, tag ==> score
book page, tag ==> score
illustration, tag ==> score [ in future]
Gamifying reader experience by presenting most relevant tags to him.
We can also assign emojis to tags (specially entities), on pattern of whatsup chat or FB post .. Emojis will also nicely capture the different emotions ..
Hi , This is an idea that occurred to me while reading the stories. (based on anecdotes) While reading
https://storyweaver.org.in/stories/131-babachya-mishya I found that Marathi translation has more tags and many of them very good, but Hindi translation has only 3 tags. So first step For stories with multiple translations, merge tags (english) from all translations .. These will form the candidate set of tags Next we can take page wise content and can use wordNet or conceptNet to find a confidence score of these tags to sentences/page. Top scoring tags can be suggested to end reader for validation while they are reading these stories on website. They can validate a tag by simple up/down method. And this can further validate.
So overall approach
We can also assign emojis to tags (specially entities), on pattern of whatsup chat or FB post .. Emojis will also nicely capture the different emotions ..