Liquefaction not summary
liquefaction instead of summary.
I've been looking into Summarizing Technology for a while now, and I had the sense that what I was looking for was not here.
Consider what it is.
The original text is large, and the summary breaks it down into smaller pieces and then picks up some of the pieces.
What are the criteria for picking up on this part of the pickup?
This standard is within the individual.
And yet, the automatic summary picks up on the "criteria shared by all".
So you only try to present to humans what you have after you pick it up.
What is important is the process of clarifying the criteria within the individual.
Letting the criteria output in the form of "keywords" and returning snippets and links that match them is search (e.g. for someone using a search engine).
Reveal the "criteria" that the individual has not yet verbalized by choosing from the options, and make a presentation based on that.
This is behavioral targeting advertising.
[Behaviorally targeted advertising - Wikipedia https://ja.m.wikipedia.org/wiki/%E8%A1%8C%E5%8B%95%E3%82%BF%E3%83%BC%E3%82%B2%E3%83%86%E3%82%A3%E3%83%B3%E3% 82%B0%E5%BA%83%E5%91%8A] #Behavioral targeting
Use keywords and other information sources used in your search.
We need a system to chew up the long crystallized text into reasonable sized fragments again.
The story is that they thought they should do that first and then build the back-end process.
Segmentation by BERT
---
This page is auto-translated from /nishio/要約ではなく液状化 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.