giladnaor/ReplyOMeter
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Initial version

Browse Source
This commit is contained in:
Gilad Naor 2024-09-09 08:34:48 -04:00
parent bcf195103c
commit 99cfc87642
2 changed files with 78 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
Design.md Normal file
View File

@ -0,0 +1,36 @@
The purpose of this document is to align on the initial design of Reply-O-Meter. Everything here is very early and likely to change.
## Overview
The following is a product-oriented view of all the stages that need to happen in the final V1 product:
1. *Data Ingestions*
Input: raw photos
Output: List of Artifacts that link the raw photo with a textual representation
a. Ingest all of the raw data files. Photos of letters, postcards, and photos.
b. Digitization: conversation of the data to a textual representation.
c. Normalization: translation of all material to one internal language (e.g. English)
d. Artifacts: creating Artifacts from joint raw material. Example: photos of person and the name/time from the photo's backside.
2. *Data Processing*
Input: List of Artifacts
Output: Graph of Entities (Person, Location, and Event)
a. Metadata (move to Ingestino?): for each Artifact, extract the Metadata on the Entities that it refers to, such as Person, Event, and Location.
b. Reconciliation: Create and/or Update existing Entities based on the information from the new Artifacts.
3. *Browser*
Input: Graph of Entities
Output: Updates to Artifacts and/or Entities
a. Feedback: User may correct any of the Ingestion or Processing steps, which will retrigger the rest of the flow.
i. Will trigger model training/tuning if relevant.
4. *Story Creator*
Input: Graph of Entities
Output: Story
a. Chat Agent: User can chat with an Author to create stories based on the known Entities.
5. *Graphic Artist*
Input: Story, Graph of Entities
Output: Graphic Novel
a. People: Given a style, create visual representation for each Person across their life.
b. Supporting Material: Gather maps of the Location and the relevant Events.
c. Create a graphic novel based on the story.

42
PR-FAQ.md Normal file
View File

@ -0,0 +1,42 @@
# The Naor Family Announces Replay-O-Meter
Bringing family stories to life
Atlit Today--September 2026--Today at the annual Jonathan birthday celebration, the Naor Family announced the release of a new software tool set that enables anyone to bring past family stories to life. Taking inspiration from Jordan Mechner's graphic memoir, [Replay](https://www.jordanmechner.com/en/books/replay/), the Replay-O-Meter makes it easy for anyone to craft stories from their own family.
"I just uploaded a bunch of old family letters and photos and Reply-O-Meter automatically identified my grandparents and their cousins. It took just a few minutes with the tool to surface some amazing stories that I never even knew about!" said Y. N., a long standing member of the family.
"I loved how I could use Reply-O-Meter to create customized graphic novels that capture the likeness of my great-grandparents. The visuals are great, but what really blew me away is how it made it easy to share great *stories*! My nephews can't get enough of these graphic novels!" says A. N., another legendary family member.
# Frequently Asked Questions
**How do I get started?**
Just scan and upload all of your old family letters, postcards, and photos. It will initially take a few hours for the software to scan, translate, and create the relevant taxonomies. Once this process is done, you will be able to browse the people and places and start crafting your stories.
**What languages are supported?**
Currently English, German, and Hebrew are supported. All of the content will be automatically translated to the desired output language.
**The software misunderstood a letter, postcard, or photograph that I uploaded. What do I do?**
In the browser, you can open any of the artifacts that you uploaded. Next to the original photo you will find the transliteration, translation, and extracted metadata. You can manually correct any mistakes in the object or handwriting recognition, the translation, and the metadata. For example, you can connect two identified personas that refer to the same person but with different names (due to marriage, immigration, etc.)
**I added more letters and there are barely any mistakes this time. How come?**
As you manually correct mistakes, the software improves its ability to read the handwriting specific to your family members. It also learns the lingual nuances specific to the time and location of the letters.
How do I create stories?
In the Story Creator, you can chat with your personal virtual author. They already know all of the people, places, and events mentioned in the material that you uploaded. You can ask it to craft a story based around a series of events or people and to use various story telling styles. Once the virtual author creates the story draft you will be able to edit and correct the narrative.
**How do I create a graphic novel?**
When the story is ready you can turn it into a graphic novel! Based on the photographs that you uploaded, the software will create a graphical representation of the people and places in the style that you choose. It will automatically adjust the age of each character to fit the time in which each event took place.
**Wow, it added maps to the story, how did it do that?**
Based on the time and location of each event described in the letters and postcards, the tool automatically pulls high quality maps from the relevant period. It also understands events that are mentioned in the letters based on publicly available data
# Internal FAQ
**That's a lot of features!**
And that is not a question. Yes, it is a lot of feature. Many can be dropped for the first version.
**Reply-O-Matic. Really?**
It's a terrible name. It's a working name. Please suggest better names. Please.
**Why now?**
Generative AI, both diffusion models and large-language-models (LLM) are enabling technologies that make this a possible product to build.