tanmayb123 / asktanmay-nlqa-system-

AskTanmay (codename: TanmayQA) is an open-source NLQA (Natural Language Question-Answering) system, developed by Tanmay Bakshi (TajyMany).

Welcome to AskTanmay!

This repository contains the complete source code for AskTanmay.

Current Versions

(Stable) 2.1 (master)

(Beta) 3.0 (TQA-3.0-beta)

---

What is AskTanmay?

AskTanmay is a Natural Language Question-Answering system (NLQA system), built in Swift, Java, and Python (though the eventual aim is to have it solely in Swift). The goal of AskTanmay is to allow developers to have a foundation to start off of, if they'd like to implement their own NLQA systems, but don't know where to start.

---

Currently Supported Answer Types

• Person (e.g. "Who is the CEO of IBM?")
• Organization/Company (e.g. "Which worldwide retail chain makes the most profit?")
• Location (e.g. "Where is the Taj Mahal?" or "In which city is the Taj Mahal?")

---

Build instructions

In order to build AskTanmay, you must have the following dependencies:

• Python 2.7.10
• Java 8
• Xcode 8.2.1 (latest stable Xcode)
• macOS

NOTE! Do not build AskTanmay from Xcode! It is not designed for this, though with a bit of modification, can run. While it will build successfully, it won't work in the later stages, as the Python and Java components have not been added. However, you can use Xcode as an IDE, and Build ("CMD+B"), not Run ("CMD+R"), to check your code, and save it.

Before building & running AskTanmay, you will need to fill out some credentials & API Keys. You need to fill out credentials in the following files:

• TanmayQA/GlobalConstants.swift
• TanmayQA_NER_FNS/NER_FNS.java (at service.setApiKey("<ALCHEMYAPI KEY>");)

Now, the build instructions:

1. Go into Terminal
2. Navigate into the directory for the project (the directory containing "TanmayQA.xcodeproj")
3. Run this bash command: sh buildTQA.sh

Done! This should compile AskTanmay and the Python & Java dependencies.

In order to run the binary, do this:

1. From that project directory, navigate to "build/Release"
2. Run AskTanmay with a question: ./TanmayQA "<QUESTION>"

---

Credits

AskTanmay's dependencies consist of the following software:

• IBM Watson java-sdk (v2.9.5)
• gson (v2.3.1)
• okhttp (v2.7.4)
• okio (v1.4.0)
• topia.termextract
• zope.interface

Thanks to the above sources for making AskTanmay possible.

License

AskTanmay (codename: TanmayQA) by Tanmay Bakshi is licensed under a Creative Commons Attribution 4.0 International License.

asktanmay-nlqa-system-'s Issues

Please give me a precompiled binary of AskTanmay for Ubuntu 18.04.1 LTS

I am unable to compile AskTanmay, Tanmay Bakshi dosen't provide precompiled binaries, I need a precompiled binary for Ubuntu 18.04.1 LTS, I'm going to criticize Tanmay Bakshi for not providing precompiled binaries, I'm tired of compiling, every project on GitHub needs a precompiled binary, in a precompiled binary, you just run and there you go, AskTanmay's compile quality is broken, please give meeeeeeeeeee a precompiled binary. I need to be able to just run and there you go that's program, that is much easier to do than compiling to most people, I have no experience with IBM Watson, also I can't find any compatible depend-ices, Come on Tanmay Bakshi, you're better than this, or I'm going to hate Tanmay Bakshi, please give me a precompiled binary for AskTanmay now now now! just for Ubuntu 18.04.1 NOW NOW NOW!, or I'm going to hate Tanmay Bakshi

Please explain global constant swift file

Hi
We know you need to get a user name and password for a Watson Service but it is not clear what services are we using. here are the issues
From the movie on youtube the ATD is using the Natural Lang. Classifier . Is that correct ?
From the movie on youtube the S-ATD is ALSO using the Natural Lang. Classifier . Is that correct ?

Can we keep the classifier_ID you provided ? because I do not see one provided by Watson
Also - why do you need the SERSE_GOOGLE_CX for ? CX is for custom search only - and I saw no API key for it

let ATD_CLASSIFIER_ID = "3a84d1x62-nlc-18693"
let ATD_CLASSIFIER_UN = "[ATD Watson UN]"
let ATD_CLASSIFIER_PW = "[ATD Watson PW]"

let SATD_CLASSIFIERS = ["GPE": "3a84d1x62-nlc-20473"]
let SATD_CLASSIFIER_UN = "[SATD Watson UN]"
let SATD_CLASSIFIER_PW = "[SATD Watson PW]"

let SERSE_RESULT_PAGES = 5
let SERSE_GOOGLE_KEY = "[Google CSE API Key]"
let SERSE_GOOGLE_CX = "[Google CSE CX]"

Swift 3

Update for Swift 3

how i can setup AskTanmay-NLQA-System into Linux Ubuntu cloud

Hello, I wants to setup AskTanmay-NLQA-System into Ubuntu 16.04 LTS. Need help @tanmayb123

More Answer Types

Implement more answer types, such as:

1. Yes/No
2. Date

CAG (Python) not working: zope missing

The "zope" library is missing - CAG is not working as a result. Fix this.

Cross-compatibility

Update to work with Ubuntu & macOS.

Clear terminal screen before beginning

In Swift 3, the system() function is not accessible. Find another way to run system("clear").

Cross Platform Capability

@tanmayb123 This looks to be an amazing idea that you have here sir! There are 2 questions though, have you thought about extending this such that you can target multiple platforms? The reason being is that currently your code is written in Xcode (I'm assuming here). However, if you could be able to have this to use other platforms (i.e. .NET possibly)? Just wondered if that idea ever came up when designing this.

Finally, speaking of making this work with .NET, and this more of a request, is there a possibility to publish a getting started guide as well? That would help with the exploration of this a lot easier! Great work again!

CAG Migration from Python to Swift

Migrate CAG (Candidate Answer Generation, or N-gram Mining) to Swift.

NER & FNS (Java) code translation to Swift

Translate NER & FNS code from Java to Swift

Port to Natural Language Understanding from AlchemyLanguage