Yewno provides AI-powered financial data through its Knowledge Graph, extracting insights from unstructured sources. It measures concept relationships with five metrics across 10,000+ securities, helping investors discover hidden connections & trends
Who are Yewno¶
Yewno provides Augmented Intelligence.
Yewno’s mission is that of extracting knowledge from an overwhelming quantity of unstructured and structured data. Our technology helps to overcome the “Information Overload” problem and to research and to understand the world in a more natural manner.It is inspired by the way humans process information from multiple sensorial channels and it leverages state-of-the-art Computational Linguistics, Network Theory, Machine Learning, as well as methods from the classical Artificial Intelligence.
What is the Yewno Concept Exposure Dataset
The Yewno Concept Exposure Data Feed leverages the Yewno Knowledge Graph to provide scores that quantify several exposure metrics from entities to key concepts as well as text data explaining or providing evidence of the connection between the concepts.
How does the core technology work?
Understanding the nature of knowledge.
At the core of our technology is the framework that extracts, processes, links and represents atomic units of knowledge – concepts – from heterogeneous data sources. A Deep Learning Network continuously “reads” high-quality sources projecting concepts into a multidimensional Conceptual Space where similarity measures along different dimensions are used to group together related concepts. In accord with prominent cognitive theories of conceptual spaces, our space allows for both geometrical, statistical and topological operations, and it permits to aggregate basic concepts into more complex representations.
Thanks to these techniques, a graph-like network, the knowledge graph, is induced and advanced tools from the field of complex networks are utilized to unfold such networks and extract inferences. Yewno’s approach explicitly addresses the temporal evolution of the knowledge network and extracts insights by analyzing not only the nature of the interconnected concepts and communities but also their temporal evolution.
Investable Universe:
This Data has global coverage across 10,000+ equity securities Nasdaq Global Index Universe, Equities, ADRs, ETFs etc
Data Update Frequency:
Daily (EOD)
Date Range
The dataset is a rolling 5 years of data
SourceType | Source Type Requested |
---|---|
Reference Date | Reference Date of Exposure |
Window | Window of Exposure |
RequestedConceptID | Concept ID Requested |
RequestedConcept | Requested Concept Name |
RequestedIsin | ISIN Requested |
ConceptID | Concept Exposed |
Concept | Concept Name Exposed |
ConceptIsin | Concept ISIN Exposed |
ConceptParentEntityName | Concept Exposposed Parent Entity Name |
ConceptParentEntityCID | Concept Exposposed Parent Concept ID |
ConceptParentEntityIsin | Concept Exposposed Parent ISIN |
Aggregate | Aggregated is a weighted linear combination of the previous scores |
Similarity | Similarity is based on how close the Company and Concept are to each other in the graph embeddings space, taking into account the network of all concepts |
Centrality | Centrality is of how central a concept is in the network of all concepts related to the company, it takes into account indirect connections as well |
Contribution | Contribution is a measure of how much the concept was co-mentioned with the company in the news or published documents relative to all the mentions of the concept. It measures how much the company contributes to the concept |
Pureplay | Pure Play is like Contribution but what percentage the concept contributes to the company, a high score would mean much of the mentions of the company is related to this concept |
- Information today is Fragmented
- Information is NOT the same as Knowledge
- It would take trillions of years and thousands of analysts to process all of the unstructure information available today
- The classical information economy is NOT SCALEABLE
Most people understand the idea of keyword extraction, and anyone who has tried this understands the drawbacks... you are looking for information on Apple computers and you get news articles about the Cider Harvest tainting your results.
To understand what Yewno do, imagine a three dimensional space where you plot 'concepts' or 'core ideas' as interconnected 'Points' / 'Planets' / 'Atoms'. Yewno calls these CONCEPTS.
Concepts can be just about anything but are most often nouns (person/company/place/idea/other) for example 'jaguar' or 'car' or 'apple' or 'phone', peoples names like 'Elon Musk' or 'Kanye West', company names like 'Tesla' or 'TSLA' or 'Apple' or 'Jaguar', ideas like 'monetary policy', events like 'Quantitative Easing' or 'Olympics', financial measures like 'Interest Rates' or 'Unemployment Numbers'.
As you can see, single words can have multiple defininations so these will exist as individual 'concepts', did you mean 'apple' the fruit or 'Apple' the company and all of these are interconnected to a greater or lesser degree.
As you can imagine, these interconnections would need more than a simple three dimensional space to map them, more like a multi-dimensional space.
The Machine Learning used by Yewno can keep track of the multi-dimensional connections and relationships between these concepts.
They process a vast corpus of high quality unstructured data feeds in real time :
- News
- Financial Documents
- Global Patents
- Government Contracts
- Clinical Trials
- Judicial Court Documents
- Corporate Filings
- Scientific Articles
Note : Customers rarely select connections across all of the feeds.
Their 'Inference Engine' tracks the relationships and the changes in those relationships over time.
Any connection can always be tracked back to the original sentence in the original source document.
And by monitoring these relationships and connections over time we can observe what is happening in the world. Detect anomolies, novelty and score the relevancy of any item.
Yewno currently (mid 2021) has ~6 million unique 'concepts', growing constantly.
Their process allows you to easily pull data relating to Jaguar Cars which are in a completely different space, with different connections when compared to the animal 'Jaguar'.
Single word, multiple meanings - Depression
Geology? Weather/Air Pressure? Economics? The 1920-21 Depression? The Great Depression? Pyschology/Depression? Physiological Depression? Each of these gets its own concept ID.
Interconnectedness : Bitcoin
'Bitcoin' is a type of 'Cryptocurrency' which affects demand for 'GPU's manufactured by 'NVIDA' symbol 'NVDA' amongst others. Yewno can detect an 'inferred connection' between Bitcoin and NVDA. It may take a large number of different source articles for the Knowledge graph to build these inter-connections.
You submit the concepts Brexit or Corona or AI or Sexual Discrimination and find all concepts (nodes/atoms) that are connected, together with their Exposure Scores.
It is like identifying a Concept Planet and being shown all other planets that orbit nearby.
Exposure Scores
Pureplay Score
Selected concept(person/company/geolocation/other) is very focussed on the target concept(person/company/geolocation/other). Example : in the 'AI' concept, a GPU company that does nothing but make TensorFlow GPUs
Contribution Score
A company with a high constribution score can have other aspects to its business but it contributes highly to the concept. Example : Again, in the 'AI' concept, Nvidia's main focus is Graphics cards for gaming but overall they contribute a lot of the core tech, news and ideas etc to AI so their score would be high. If the user pulled connections only from the 'patents' source then Nvidia would have an extremely high contribution score under the concept 'AI' as they have probably submitted more AI related patents than any other company.
Centrality Score
Companies with high scores are closely aligned with the concept (company/person/geolocation/idea/other) submitted. Example : For Tesla highly correlated concepts would be the symbol TSLA, Elon Musk, Automotive Industry, gigafactory, Lithium Batteries, Austin, Berlin etc
Similarity Score
Angular distance from the submitted concept to the returned concepts. Example : The concept "Apple Iphone" would be very similar to the concepts "Blackberry" or "Android Phone" but not at all similar to "Apple" (Fruit).
Aggregate
A linear combination of the other exposure scores.
Dataset Names
Yewno Sentiment, Concept and Snippets
-
yewno_sentiment_cid
-
yewno_sentiment
-
yewno_concept_cid
-
yewno_concept_keyword
-
yewno_concept_keyword_company
-
yewno_concept_keyword_person
-
yewno_concept_keyword_place
-
yewno_concept_keyword_other
-
yewno_snippets
Yewno Scores for News, Official Filings, Clinical Trials, Patents and Transcripts for 2, 15, 180 and 365 days plus Metadata
-
yewno_score_news_2d
-
yewno_score_news_15d
-
yewno_score_news_365d
-
yewno_score_news_180d
-
yewno_metadata_news
-
yewno_score_cid_news_2d
-
yewno_score_cid_news_15d
-
yewno_score_cid_news_180d
-
yewno_score_cid_news_365d
-
yewno_metadata_cid_news
-
yewno_score_officialfilings_2d
-
yewno_score_officialfilings_15d
-
yewno_score_officialfilings_180d
-
yewno_score_officialfilings_365d
-
yewno_metadata_officialfilings
-
yewno_score_cid_officialfilings_2d
-
yewno_score_cid_officialfilings_15d
-
yewno_score_cid_officialfilings_180d
-
yewno_score_cid_officialfilings_365d
-
yewno_metadata_cid_officialfilings
-
yewno_score_clinicaltrials_2d
-
yewno_score_clinicaltrials_15d
-
yewno_score_clinicaltrials_180d
-
yewno_score_clinicaltrials_365d
-
yewno_metadata_clinicaltrials
-
yewno_score_cid_clinicaltrials_2d
-
yewno_score_cid_clinicaltrials_15d
-
yewno_score_cid_clinicaltrials_180d
-
yewno_score_cid_clinicaltrials_365d
-
yewno_metadata_cid_clinicaltrials
-
yewno_score_patents_2d
-
yewno_score_patents_15d
-
yewno_score_patents_180d
-
yewno_score_patents_365d
-
yewno_metadata_patents
-
yewno_score_cid_patents_2d
-
yewno_score_cid_patents_15d
-
yewno_score_cid_patents_180d
-
yewno_score_cid_patents_365d
-
yewno_metadata_cid_patents
-
yewno_score_transcripts_2d
-
yewno_score_transcripts_15d
-
yewno_score_transcripts_180d
-
yewno_score_transcripts_365d
-
yewno_metadata_transcripts
-
yewno_score_cid_transcripts_2d
-
yewno_score_cid_transcripts_15d
-
yewno_score_cid_transcripts_180d
-
yewno_score_cid_transcripts_365d
-
yewno_metadata_cid_transcripts