Title: Presentazione Expert System
1Semantic Intelligence The Enabler of Mobile
Search
May 23 2007, San Jose - CA
2Introduction to Mobile Search
- The mobile phone is the most ubiquitous
communication tool, used for a variety
of activities - Voice Communications Messaging - Information
Gathering Entertainment Shopping Commerce
Much More
- The mobile phone is a new and unique medium
because it occupies a distinctly different place
in consumers lives than a PC, TV, or any other
commonly used tool.
- Given this unique nature there are some unique
customer experience heuristics that should be
considered when designing products and services
for the mobile channel.
3The Truth About Mobile Search
- Mobile search is NOT simply targeted at the
personalization and entertainment uses,
mobile search also addresses several utility
functions.
- Mobile phones accessing the Internet have
several disadvantages over PCs - Tiny screens with limited capability, unsuited
to show a lot of detailed information and
numerous search results. - Compact number pads and small keyboard
- Different input methods, display screens,
browsers, operating systems, and user interfaces
- Mobile devices do not conform to any standards.
- Thus Mobile search tends to look for an
experience that offers answers and
actionable results, rather than links that lead
to more searching.
4- Introduction to Semantic Intelligence
Semantic Intelligence is an advanced way to
transform unstructured information into
structured data that can be managed with standard
databases.
5Introduction to Semantic Intelligence
- Recognizing the meaning of the words is
fundamental to understanding the content of any
text. - Semantic Intelligence enable the understanding
of the real meaning of the words and expressions
used in the text, thanks to - Deep analysis
- Recognition of the context
- Richness and completeness of information
contained in its semantic net (Sensigrafo more
than 350.000 concepts and 1 million relations) - This works at a real conceptual level, evolving
from the bag of words to a set of abstract and
powerful concepts
6- Introduction to Semantic Intelligence
- Going deep means recognizing the inner structure
of the text and the relations between the
elements, and being able to manage the content
with much higher precision and relevance. - For example, in a sentence like this
The sandwich that John gave to me was very tasty
- The system understands that it is the sandwich
that is tasty (and not John even if it is nearer
in the text), that John gave it to me and that
the main clause is The sandwich was very tasty.
7- Introduction to Semantic Intelligence
- Working on the semantic part of the content, the
system can easily understand that this sentence
A fast car that eats gas
- Has many words in common with this sentence
A car goes fast when we step on the gas
- But the meaning of gas (and of the whole
sentence) is completely different. - In the first one, we are talking of gasoline
while in the second of the accelerator.
8Semantic Intelligence Mobile Search
Semantic Search addresses the limitations of
keyword search in the mobile space by integrating
content and producing accurate results to natural
language queries.
- Integrates structured or unstructured content
from an entitys knowledge base. - Enables users to ask questions receive
specific, accurate answers! - No need to surf through slow links or get
faulty info through keyword search.
9Semantic Intelligence How It Works
- Deep linguistic analysis
- of the natural language based query to really
understand the intent of the user AND - of the content (if unstructured) to really
understand what is available in the knowledge
base. - The system recognizes the inner structure of the
text and the relations between the elements, and
can manage the content with high precision
and relevance, such as - When is the next flight to NY? OR At what time
is the next plane to JFK ? - What did the As do last night? OR What was
the score of the As game OR Did the As win
last night? -
- The system understands that the questions have
the same meaning and that in both cases specific
information is requested.
10Semantic Intelligence How It Works
- Structured Content
- The natural language query is translated into a
DB query that is used to access the information
available. - The semantic analysis of the question in input
enables an efficient creation of the right DB
Boolean query.
Ex At What time is the next flight to
NY? Query time Now next flight first
flight after Now from current location to
New York
11Semantic Intelligence How It Works
- Unstructured Content
- Content is analyzed semantically to identify
Atoms of knowledge - Example from Wikipedia
- - Atom 1 The African and Asian elephants are
separate species. - - Atom 2 African elephants, at up to 4 m
(13'1") tall and weighing 7500 kg, (8.27
tons) are usually larger than the Asian species
and they have bigger ears. - The embedded validated and structured knowledge
can be leveraged immediately, requiring an
integration for only very specific domains - Conceptual map of the content is created
- Question is analyzed to understand what the user
is asking for requested - Example How tall are the elephants?
- The system automatically links the info requested
and the atom of knowledge (including the
answer) using the semantic information extracted
from the analysis of the question and the results
of the analysis of the answer.
12- For a complex technology, speed of execution is
fundamental What would have happened if Google
hadnt implemented PageRank algorithm in
such a fast way? - COGITO has been designed and implemented with
an industrial approach that performs one or two
orders of magnitude faster than anything else
(50KBps of text analyzed). - The speed of analysis translates into higher
precision because the system can analyze a
high number of options and choose the
best possible one. - Our customer base proves that COGITO can be
used in all projects and
industries without limits on scope, volume
or performance.
13- Customer Service
- Extremely relevant, very real and pressing need
in the mobile space. - Large companies are spending millions yearly on
answering customer queries, this type of semantic
tool greatly reduces cost and increases
efficiency! - Deliver seemingly endless amounts of info
through simplest query answer format ideal
for mobile phone. - Allows customers to ask questions such as
- How does it work?
- What is my account balance?
- Is there a widget of any spec in stock?
14Usage Benefits
- Directory Assistance - 411
- Current 411 DBs and search are very limited by
rigid structure and menu navigation, resulting in
much less data being easily accessible from a
mobile phone. - Deliver much greater amounts of relevant info
and data points through simplest semantic query
answer format. - Customers can easily input content to the DB
make each result more detailed and relevant - Allowing customers to interact simply and
without boundaries - Q Where is the best NY style pizza place near
Russian Hill that delivers with good salads and
open until midnight? - A Amicis Pizza, 415-885-4500, 2200 Lombard
Street, 94123
15Usage Benefits
- Extending Content
- Content Publishers have a difficult time making
their vast amounts of info easily extendible via
mobile, where minimal user interface hurts their
users ability to navigate the content. - Now content publishers can deliver the exact
desired info to mobile users through easy
semantic query and answer. - Encyclopaedia Britannica users can ask questions
such as - When was George Washington born?
- Who was George Washingtons wife?
- What famous river did George Washington lead the
troops across? - Instead of searching for George Washington
then going through all articles, the exact info
desired is delivered!
16More Examples
Who coached the Jets in 2003?
Does Mr. Doe live in Pomfret, CT?
How do I send email with my Blackberry?
Who sings Imagine?