7

Yes, it is possible to combine probabilistic / bayesian reasoning and a traditional "knowledgebase". And some work along those lines has been done. See, for example, ProbLog ("Probabilistic Prolog") which combines logic programming and probabilistic elements. See: https://dtai.cs.kuleuven.be/problog/tutorial/mpe/01_bn.html Another project to look at ...


5

You can use a component library which can help you to implement Natural language query builder in your application( the question part ) called Open Natural Language Processing Package , so you can definitely develop a module, by using existing modules of OpenNLP such as entity extraction, chunking and parsing. According to wikipedia source ; it points ...


4

For vision tasks, neural network models almost always include a number of layers that pool and convolute. The convolutions, in particular, are very useful - they can make the model generalize better to inputs and maintain performance when inputs have undergone certain linear transformations (e.g. some scaling or a translation along the x-axis). These ...


2

You can use Google https://encrypted.google.com/search?hl=en&q=when%20was%20Einstein%20born and parse the response. Wolfram ALPHA is another candidate. http://m.wolframalpha.com/input/?i=what+year+was+Einstein+born&x=0&y=0 You can parse the returned html and see "Result:" div.


2

If you are talking about "generating" in the sense of generative models , it is pretty tough. since we are still far beyond understanding the actual structure of question-answering. And even state of the art methods for question answering are also not able to score well on datasets like babi , mostly 16 out of 20 tasks can be solved.


2

There was a lot of work on this topic at UT Austin, which has now migrated to the Alan Institute. There is no off-the-shelf software that will answer your question (if there was, DARPA would stop funding its development!), but you can read about the latest development in a number of recent papers. This paper (Seo et al. EMNLP 2015) discusses the techniques ...


2

Sophia ,first , has all the questions and corresponding answers preprogrammed. It is a system which is a hybrid of Bayes Text classification and decision trees. It may consist of a speak recognizer which converts the question into a string. This string then travels into a algorithm which gets a suitable answer for it. The answer is then spoken by the ...


2

In QA, it's computed over the individual words in the prediction against those in the True Answer. The number of shared words between the prediction and the truth is the basis of the F1 score: precision is the ratio of the number of shared words to the total number of words in the prediction, and recall is the ratio of the number of shared words to the total ...


1

What you want to look for is called anaphora resolution. You basically keep a record of the past conversation and try and find an antecedent for any occurrences of it, he/she, her/his, etc. You probably want to have a pre-processing step where you substitute the antecedent before passing the input sentence on to the agent.


1

The answer is yes but 'lightweight' will require a 'lightweight' model. Your application for 'domain one' is called open domain question answering (ODQA). Here is a demonstration of ODQA using BERT: https://www.pragnakalp.com/demos/BERT-NLP-QnA-Demo/ Your application for 'domain two' is a little different. It is about learning sequences from sequences. ...


1

Well this is a relatively new problem very tied to Question Answering. One of the recent systems is EUCLID that can answer those type of question the public Dolphin algebra question set by using a tree transducer cascade approach. This paper details the proposed model Hopkins, M., Petrescu-Prahova, C., Levin, R., Le Bras, R., Herrasti, A., & Joshi, V. (...


1

This repository maintained by Facebook AI Research talks about how they went about generating QA from stories. In essence, they try to simulate how a reader reads a story. They also keep track about the knowledge the reader is assimilating when reading. Then they frame a question based on the knowledge assimilated, in order to asses if the reader can ...


1

I think the issue here is that the chatbots you're using aren't very good at "short-term memory". What I mean by is that the bots construct responses that are slowly and incrementally tuned according to the overall usage of the chat bot, from every user. The bots are responding to each message based on how a new user would expect them to. As Alan1 notes, "...


1

I have been reading and reading, and found answers to almost all my questions. I am sticking to Earley algorithm, given that it offers a dynamic programming approach (CKY does the same). Both algorithms are chart parsing algorithms. Earley is a context-free, top-down parsing algorithm, which makes it a goal-driven algorithm. From start symbol down. ...


1

You could use dbPedia and/or wikidata. I think Wikidata supports SPARQL now, but don't quote me on that. dbPedia definitely supports SPARQL. If you're not interested in writing SPARQL queries by hand, you could use something like Quepy. In fact, the Quepy demo demonstrates doing natural language queries against Freebase and/or dbPedia. You could ...


Only top voted, non community-wiki answers of a minimum length are eligible