Upload
alicia-harapko
View
63
Download
1
Embed Size (px)
Citation preview
Bryan BellExecutive Vice [email protected]: @bellbryan
Semantics Helps Connect the Dots
Enterprise Data World ConferenceSan Diego, CAApril 19, 2016
Session topics: 4 topics
1. What is semantics and What is a semantic technology platform?
Semantics Helps Connect the Dots
2. How to use semantic technology to:
• Leverage internal and external information• Co-mingle structured and unstructured content and establish connections• Promote knowledge sharing
3. How semantic technology can improve traditional data management methods:
• Dynamic metadata enrichment & content tagging• Automatically create contextually correct metadata• Automated content categorization & classification • Mapping with geo-tagging• Cluster content with like content
4. How semantic technology can provide a window into how people, places, things and events come together. Identify opportunities to support business objectives.
examples
“I can’t find what I need to do my job!”
McKinsey Report (2013): Most common reasons
information can not be found— Lack of structure in content
• no (read: metadata)• poor quality mtdata• inconsistent meta-data• inaccurate meatdata
— Difficult to access (multiple data silos)
— Got tired of looking / gave up
— Don’t know how to ask for it.
Result: Phone a friend or recreate content
We all spend too much time searching.
McKinsey Report: May 2013
— Employees spend 1.86 hours every day— 9.3 hours per week, on average— Searching and gathering information.
Put another way: — Business hires 5 employees — While 4 show up to work… — …the 5th is off searching for answers
Money wasted:100 employees (20 are searching)$80k x 20 employees = $1,600,000/year
LOOKING FOR
1000 employees (200 are searching)$16,000,000 annually in lost productivity
Semantics Helps Connect the Dots
Problem:• 3 V’s (volume, variety, velocity)
Internal and external information comes at us faster than we can keep up with.
Today’s topic:• Ways to improve: enterprise search, content navigation, information discovery &
knowledge sharing.
Comparison of available options: • Key words vs. Statistics vs. Shallow Linguistics vs. Linguistics + Semantics
• Business expectations to address the problem to be able to quickly locate and leverage corporate knowledge have not changed.
Improve: Enterprise search, content navigation, information discovery, knowledge sharing
1995 – 21 yearsWhat I have learned is that customers and vendors
(generally speaking) are working to achieve the same result.
Semantics Helps Connect the Dots
Organize content
Structure in a consistent way• Make it findable at the right moment• Make it related to the problem at hand• Reusable & shareable going forward
Customers point of view Difficult to understand the options.
Difficult to differentiate the approaches available in the market.
Key words – language ambiguity is a primary challengeo jaguar vs. jaguar vs. jaguar
Industry approaches to address language ambiguity
Statistics – “black box”o How does it work?o Representation of words look the same, but may have different meaningo Can not be modifiedo Very difficult (nearly impossible) to correct mistakes.
Shallow linguisticso Identifies the sentence elements (noun, verbs, adjectives, etc.) o But does not specify their role in the main sentence. (logical analysis)
Linguistic analysis combined with Semantic disambiguationo To better understand word sense. To remove word ambiguity.
The purpose: Establish word context.
Morphological analysis word forms dog, dog-catcher, doggy bag
Grammatical analysis parts of speech "There are 40 rows in the table." (noun)
"She rows 5 times a week." (verb)
Logical analysisword
relationships"The car I bought, to replace my Chrysler,
stinks."
Semantic analysis word context “stock”
Linguistics combined with SemanticsLinguistic analysis combined with semantic disambiguation
stock“I bought 10,000 shares of stock in Apple.”
“I have 10,000 apples in stock.”
“We do not have that gun stock.”
“We do not have that gun in stock.”
“I used chicken broth for my soup stock.”
Establishing word context
Grammatical analysisLogical analysis
Establishing word context
Word relationships / compound terms / multi-word concepts
Semantic network(language ontology)
Semantic Network (language ontology)
Stock48 versions3 versions
Over 50 relationship
options
Morphological, grammatical and logical analysis combined with semantic analysis.
Semantics (word disambiguation) Contextual analysis
Business Use Cases
Dynamically created metadata Contextually correct metadata
Providing the structure (metadata)Making content reusable & findable
• Challenge: Difficult to persuade employees to add quality metadata.
• Consistency: employee to employee or department to department?
• Does it adhere to the corporate metadata model?
• Internal or external documents• news media• Competitors web sites• e-mail• chat sessions• Networked shared drives• Social media
Primary requirement:
- Electronically accessible - Converted to .txt
Semantic networkDynamically created, contextually correct metadata
www.intelligenceapi.com
Dynamically created, contextually correct metadata
Step 1: All documents areanalyzed using linguistics combined with semantic reasoning.
• Combine many desperate data sources• Taxonomy creation & management
• Organize content in a precise way
Connecting related content
Identifying networks of people, places and events
Discover the undiscovered
New objective: FindTell me what I do not know?Discover new information.
Oil & Gas
Federal agencies
Social Media & Sentiment Analysis
Pharmaceutical &Life SciencesMedia & Publishing
Legal analysisMobile &
telecommunication
This is not a industry specific problem.It is a language ambiguity problem.
Insurance, Financial Services
Semantics Helps Connect the DotsVendor questions
1. Is it key word, statistics, lingustics, semantics or a combination (combined approach)?
2. How is language ambiguity addressed?
3. How is word context established?
4. Is metadata manual or dynamic?
5. Request «live» demonstrations on real / current content.
Ideal situation – customer content