Weitere ähnliche Inhalte Ähnlich wie Semantic based Enterprise Search Solution in Networking Domain (20) Kürzlich hochgeladen (20) Semantic based Enterprise Search Solution in Networking Domain2. 2 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
SCENARIOS
I want to see facts
from different
sources describing
EX4200
I want to more
about EX4200
about its chassis
specifications
I want to know
power supply
details of EX4200
switches
I want to know
more about
Juniper’s Security
solutions
I want to know
Juniper’s Data
center offerings
I want to see
customer’s
feedback about
Junos
3. 3 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
1. Find Information faster
Provide search assistants
2. Reveal hidden information
Enrich the search index with background
knowledge
3. Find more specific information
Query the semantic web
4. Find linked information
Integrate data sources
4. 4 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
FIND INFORMATION FASTER
To Provide a powerful auto-complete based on past searches
To have a enterprise vocabulary to assist in auto-complete
I can’t
remember
how to spell
the search
term
5. 5 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
FIND INFORMATION FASTER
High Performance Networking Search
I can’t
remember
exactly what I
was looking
for
6. 6 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
FIND INFORMATION FASTER WITH RELATED SEARCH
TERM
EX4200 Search
7. 7 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
REVEAL HIDDEN INFORMATION WITH QUERY
EXPANSION
VLAN SearchOR ”VLAN Configuration"
Ontology
VLAN
VLAN Configuration
alternative Label
preferred Label
8. 8 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
FIND LINKED INFORMATION
IX Content
KB
TechNotes
J-Net
Technical
Bulletin
EX4200
9. 9 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
WHY ENTITIES IMPORTANT
Entities are new driver.
Entities are generated relational mapping that uncovers the
association between different data points.
Entities becomes trusted points around which other data
revolves.
10. 10 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
THE ROLE OF ONTOLOGY IN SEMANTIC SEARCH
Junos
12.2 R1
Junos 12.2
JunosSoftware
Sep 05
2012
ACX
1000
ACX
2000
M7i
M10i
MX5
MX10
T320
T640
EX2200
EX4200
ACX
M
Series
MX
T
Series
EX
Series
Routing
Switching
Hardware
Released on
Belongs to
Supported on
Belongs To
Belongs To
Belongs To
11. 11 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
HOW TO EXTRACT ENTITIES
Top level view, how entities are defined.
We have data sources from iX contents, KB, Technets, J-Net
forums, Technical Bulletins etc…
Through entity detection and raw relation detection raw text
extracted from web pages in unstructured data format at one end
becomes a responsive, refined entity that can provide an intelligent
answer in semantic search.
12. 12 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
ONTOLOGY A CENTRAL POINT TO CONTROL
Labels and Query Expansion
Faceted Search
Refine Search Mechanism
Metadata Integration
Search Services
Search Application
Collector
Semantic
Indexer
Document
Index
Ontology
Manager
Extractor
HTML
Ontology
DB
13. 13 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
PROBLEM WITH CURRENT SEARCH
Ambiguity and association in natural language
Search is based on text
Polysemy: Words often have a multitude of meanings and different
types of usage (more urgent for very heterogeneous collections).
Synonymy: Different terms may have an identical or a similar
meaning (weaker: words indicating the same topic).
14. 14 Copyright © 2013 Juniper Networks, Inc. www.juniper.net
LATENT SEMANTIC INDEXING
Latent Semantic Indexing (LSI) means analyzing documents to find the
underlying/latent meaning/semantics or concepts of those documents.
The fundamental difficulty in finding relevant documents from search words is that
what we really want is to compare the meanings or concepts behind the words.
LSA attempts to solve this problem by mapping both words and documents into
a "concept" space and doing the comparison in this space.
LSI overcomes two of the most problematic constraints of Boolean keyword
queries:
multiple words that have similar meanings (synonymy)
words that have more than one meaning (polysemy).
Text does not need to be in sentence form for LSI to be effective. It can work
with lists, free-form notes, email, web content, etc.
LSI is also used to perform automated document categorization and clustering. In
fact, several experiments have demonstrated that there are a number of
correlations between the way LSI and humans process and categorize text.