I am new to machine learning, I want to extract event triple from news headlines. But there are many news headlines without event triples, only containing subject and predicate. How shall I deal with these kind of sentences?
Association rule mining is considered to be an old technique of AI. Rules are mined on statistical support. How can deep learning be applied to this? What are approaches for structured data (in a graph format like XML)? XML documents are structured by tags. My goal is to extract a rule that says that tag x is often combined with tag y and z. Then, I later want to apply these rules and if a tag y and z is …
I'm trying to use PyTorch BigGraph pre-trained embeddings of Wikidata items for disambiguation. The problem is that the results I am getting by using dot (or cosine) similarity are not great. For example, the similarity between the Python programming language and the snake with the same name is greater than between Python and Django. Does anybody know if there is a Wikidata embedding that results in better similarities? The only alternative I've found is Webmembedder embeddings but they are incomplete. …
Suppose I have 2 sentences that I have created using the below code WITH split(tolower("His dog eats turkey on Tuesday")," ") as text Unwind range(0,size(text)-2) AS i MERGE (w1:Word {name: text[i]}) MERGE (w2:Word {name: text[i+1]}) MERGE (w1)-[:NEXT]->(w2) RETURN w1, w2 WITH split(tolower("My cat eats fish on Saturdays")," ") as text Unwind range(0,size(text)-2) as i MERGE (w1:Word {name: text[i]}) MERGE (w2:Word {name: text[i+1]}) MERGE (w1)-[:NEXT]->(w2) RETURN w1, w2 The Graph looks like something. Now given this what would be a good …
I am new to machine learning and want to classify sentences using ontologies (taxonomies/ knowledge graphs) and supervised learning methods (I have an annotated training dataset). My question is how to use the ontology for this task? Is the following method correct? I will first perform the tokenization, stemming and stop word removal (pre-processing). Then, I will search for each term in the ontology and after finding them, I will add their related hierarchy to an array or vector for …
Little background on Work : I am working with ontologies and for my usecase I have to apply random walk on the ontology nodes/entities. In order to do the same I have written one function - that given a node it will output all its immediate neighbour nodes. Actual Problem : Recently I came across an ontology which along with normal nodes in the graph also has "union of other nodes" as an entity. But while going over the triple's …
I have a symptom-disease network that consists of four attributes: symptom, disease, co-occurrence and TF-IDF. I'm considering the TF-IDF attribute as the weight of my network edges and symptom and disease attributes as nodes of my network. But I think there is a problem that should be addressed. here are some records of my dataset: MeSH Symptom Term MeSH Disease Term PubMed occurrence TFIDF score "Aging, Premature" Scoliosis 1 3.46455149 "Aging, Premature" HIV Infections 3 10.39365447 Weight Loss Scoliosis 1 …
I’m a CS undergrad trying to make my way into NLP Research. For some time, I have been wanting to incorporate "everyday commonsense reasoning" within the existing state-of-the-art Language Models; i.e. to make their generated output more reasonable and in coherence with our practical world. Although there do exist some commonsense knowledge bases like ConceptNet (2018), ATOMIC (2019), OpenMind CommonSense (MIT), Cyc (1984), etc., they exist in form of knowledge graphs, ontology, and taxonomies. My question is, how can I …
By Commonsense knowledge I mean a KB like ConceptNet vs. Encyclopedic KB like DBPedia. Which one has potentially more facts? I mean whether the facts of instances is more or facts about concepts in commonsense?
I am experimenting with Fluent Editor, and I have trouble expressing something. If one creates the example project IT Infrastructure Advanced, in 7_Incidents.encnl there is this statement: Power-Node-3 is-not affected by an incident., which explicitly states, that something is not affected, and anything inferred for the power nodes (and thus all other entities) is based on the existence of this statement. This is bad, as incidents can affect things at multiple levels, and if there is a need to explicitly …
I have built an item-item recommender model using (Graph Convolutional Network (GCN) for an E-commerce website. Could you please help me with the maintenance of the model. How often should I retrain the model after deployment?( 1 month, 2 months) What are the main metrics which I should track after deployment to evaluate the model? Any suggestions related to Train/valid/test split ratios (such as 70:15:15 or 60:20:20) Any materials related to model maintenance/recommder systems evaluation post deployment would be of …
I want to create a neural network that takes as an input a knolwedge subgraph(different types of nodes and different types of edges) to predict some properties. For instance an input in the graph can be : :Mike :likes :chocolate; :studies :Biology. Assuming we have n nodes and r relation types. How can I provide this as an input in a neural network and how will the architecture of the neural network look like.