Basit Anwer
commited on
Commit
·
4bee647
1
Parent(s):
7174068
Updated Readme.md
Browse files
README.md
CHANGED
@@ -14,22 +14,22 @@ mechanism, which gets relevant information from ZIR Semantic Search.
|
|
14 |
|
15 |
# Table of contents
|
16 |
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
|
26 |
## Rasa Bot & Zir
|
27 |
|
28 |
[Rasa](https://rasa.com/) is the leading conversational AI platform, for
|
29 |
-
personalized conversations at scale. Developers provide the
|
30 |
-
expect the end customer to ask and Rasa bot, using AI, predicts and matches
|
31 |
what the customer intended to ask. With this information at hand, developers can
|
32 |
-
easily match what the bot should respond.
|
33 |
|
34 |
This is all well unless when the customer asks a valid question the developer
|
35 |
did not expect, for such a scenario Rasa provides a
|
@@ -41,13 +41,13 @@ did not expect, for such a scenario Rasa provides a
|
|
41 |
> option to either respond with a default message or attempt to disambiguate the
|
42 |
> user input.
|
43 |
|
44 |
-
Using the mechanism the bot can ask a fallback action and search through the
|
45 |
-
uploaded in [Zir-AI](https://zir-ai.com) with the help of
|
46 |
[custom actions](https://rasa.com/docs/rasa/custom-actions). If the reviews have
|
47 |
-
the
|
48 |
-
|
49 |
|
50 |
-
The
|
51 |
|
52 |
## Setup & Play
|
53 |
|
@@ -59,7 +59,7 @@ the dependencies rasa requires.
|
|
59 |
|
60 |
Follow the
|
61 |
[rasa installation instrcutions](https://rasa.com/docs/rasa/installation) or
|
62 |
-
copy
|
63 |
|
64 |
```bash
|
65 |
pip3 install -U pip
|
@@ -83,7 +83,7 @@ Python Path : /bin/python3
|
|
83 |
|
84 |
### Set-up repo
|
85 |
|
86 |
-
To install
|
87 |
from requirements.txt
|
88 |
|
89 |
```bash
|
@@ -100,14 +100,14 @@ rasa train
|
|
100 |
|
101 |
### Running the demo
|
102 |
|
103 |
-
While running the model, you are
|
104 |
with the rasa bot. So in one terminal run
|
105 |
|
106 |
```bash
|
107 |
rasa run actions
|
108 |
```
|
109 |
|
110 |
-
In another terminal you can either play with the bot in your `shell` or on the
|
111 |
browser.
|
112 |
|
113 |
To talk to the bot in shell run
|
@@ -131,20 +131,20 @@ Use the following information to customize the bot and the source of data.
|
|
131 |
### Data
|
132 |
|
133 |
Data for the bot is sourced from [OpinRank](https://github.com/kavgan/OpinRank/)
|
134 |
-
and reviews for Hotel Jumeirah
|
135 |
parsed from https://github.com/amin3141/zir-souffle. Use `hotels.py` to generate
|
136 |
the output `json` and the `reviews.db`.
|
137 |
|
138 |
-
Create a corpus on [Zir AI](http://zir-ai.com) and upload all the `
|
139 |
-
Create an
|
140 |
|
141 |
-
-
|
142 |
- corpus id
|
143 |
-
- customer
|
144 |
|
145 |
### Rasa Custom Action
|
146 |
|
147 |
-
Use the above
|
148 |
`actions/actions.py` and you're good to go.
|
149 |
|
150 |
```python
|
|
|
14 |
|
15 |
# Table of contents
|
16 |
|
17 |
+
- [Rasa Bot & Zir](#rasa-bot--zir)
|
18 |
+
- [Setup & Play](#setup--play)
|
19 |
+
- [Set-up Rasa](#set-up-rasa)
|
20 |
+
- [Set-up repo](#set-up-repo)
|
21 |
+
- [Running the demo](#running-the-demo)
|
22 |
+
- [Customizing the bot](#customizing-the-bot)
|
23 |
+
- [Data](#data)
|
24 |
+
- [Rasa Custom Action](#rasa-custom-action)
|
25 |
|
26 |
## Rasa Bot & Zir
|
27 |
|
28 |
[Rasa](https://rasa.com/) is the leading conversational AI platform, for
|
29 |
+
personalized conversations at scale. Developers provide the questions they
|
30 |
+
expect the end customer to ask and the Rasa bot, using AI, predicts and matches
|
31 |
what the customer intended to ask. With this information at hand, developers can
|
32 |
+
easily match what the bot should respond to.
|
33 |
|
34 |
This is all well unless when the customer asks a valid question the developer
|
35 |
did not expect, for such a scenario Rasa provides a
|
|
|
41 |
> option to either respond with a default message or attempt to disambiguate the
|
42 |
> user input.
|
43 |
|
44 |
+
Using the mechanism the bot can ask a fallback action and search through the
|
45 |
+
reviews uploaded in [Zir-AI](https://zir-ai.com) with the help of
|
46 |
[custom actions](https://rasa.com/docs/rasa/custom-actions). If the reviews have
|
47 |
+
the relevant information is shown as a result otherwise if the confidence of the
|
48 |
+
question to reviews available is low, we fallback to a generic statement.
|
49 |
|
50 |
+
The results sum up to be this:
|
51 |
|
52 |
## Setup & Play
|
53 |
|
|
|
59 |
|
60 |
Follow the
|
61 |
[rasa installation instrcutions](https://rasa.com/docs/rasa/installation) or
|
62 |
+
copy-paste the following commands in terminal
|
63 |
|
64 |
```bash
|
65 |
pip3 install -U pip
|
|
|
83 |
|
84 |
### Set-up repo
|
85 |
|
86 |
+
To install dependencies for this repository to work install python dependencies
|
87 |
from requirements.txt
|
88 |
|
89 |
```bash
|
|
|
100 |
|
101 |
### Running the demo
|
102 |
|
103 |
+
While running the model, you are required to run the rasa action server along
|
104 |
with the rasa bot. So in one terminal run
|
105 |
|
106 |
```bash
|
107 |
rasa run actions
|
108 |
```
|
109 |
|
110 |
+
In another terminal, you can either play with the bot in your `shell` or on the
|
111 |
browser.
|
112 |
|
113 |
To talk to the bot in shell run
|
|
|
131 |
### Data
|
132 |
|
133 |
Data for the bot is sourced from [OpinRank](https://github.com/kavgan/OpinRank/)
|
134 |
+
and reviews for Hotel Jumeirah are specifically used for this bot. Reviews are
|
135 |
parsed from https://github.com/amin3141/zir-souffle. Use `hotels.py` to generate
|
136 |
the output `json` and the `reviews.db`.
|
137 |
|
138 |
+
Create a corpus on [Zir AI](http://zir-ai.com) and upload all the `JSON files.
|
139 |
+
Create an API key for access and note down the following information
|
140 |
|
141 |
+
- API key
|
142 |
- corpus id
|
143 |
+
- customer-id
|
144 |
|
145 |
### Rasa Custom Action
|
146 |
|
147 |
+
Use the above-collected information and replace the information in
|
148 |
`actions/actions.py` and you're good to go.
|
149 |
|
150 |
```python
|