![Intrinsic semantic (LWO) vs.:
Full-fledged ontologies [Supervised learning]
- Works only with detailed models
- Language dependent,
- no incrementally updating
Sharing / reuse of ontologies [limited possibilities]
- Based on models and reservations about the quality
- Language dependent
- no incrementally updating
Folksonomies [WEB 2.0 / semantic WEB]
- No controlled quality or validation
- Often incomplete or not existent, Language dependent
- no incrementally updating
ai-one
© ai-one inc. 2011
inc. 2010](https://image.slidesharecdn.com/02aionecontentanalyticsbusinesscases-13371622282143-phpapp02-120516045855-phpapp02/85/02-ai-one-content-analytics-business-cases-12-320.jpg)
02 ai-one - content analytics business cases
- 1. Business cases for content analytics with ai-one biologically inspired intelligence © ai-one inc. 2012 ai-one™
- 2. Biologically Inspired Intelligence logic creativity © ai-one inc. 2012
- 3. The secret of content analytics … with only a few ai-one commands it is easy to build and use semantics in language ! Successionaly we show some sample concepts how to use the ai-one approach (commands) in daily business cases. © ai-one inc. 2012
- 4. Make sense of a text i.e. in: E-Mail, Article, Feed, Tweet, Story etc… Recognize the sense and meaning within a text corpus means to identify the semantically most important words which build the content Use of the : KeyWordCommand © ai-one inc. 2012
- 5. Make sense of a text i.e. in: E-Mail, Article, Feed, Tweet, Story etc… Recognize the sense and meaning within a text corpus means to identify the semantically most important words which build the content Extract the most important words in a text corpus and form the Light Weight Ontology (LWO) and from a digest of the meaning. That is the condensed summary of the sense and meaning of a text. Now ai-one or other matcher can classify and sort the text. This is the ai-Fingerprint. With the for example 7 words this text is define in its sense Use of the : KeyWordCommand © ai-one inc. 2012
- 6. Find associative, semantic relations i.e. in: E-Mail, Article, Feed, Tweet, Story etc… Its like brainstorming, what has to-do with what, or which word is semantically connected with which word. Find the associative and semantic relations trough a whole big text or whole data base. Find patterns we did not know they exist! WORD This commands starts with one or multiple words WORD and searches for the semantically relations. WORD Find patterns of relations in whole text corpus. WORD WORD Validate the importance of a connection WORD WORD between two or multiple words Detect association bridges between two words. Display of semantic chains.! WORD WORD WORD Use of the : AssoAnalysCommand © ai-one inc. 2012
- 7. Find syntax patterns i.e. E-Mail, Article, Feed, Tweet, Story etc… One additional challenge is the spelling. Users very often miss spell words. Therefore we also search for syntax patterns in order to verify the words. The Phonetic pattern recognition on syntax is very helpfully to identify similar • Maier • Meyer words, spell errors and artificially re- • Mair • Peyer designed words. • Paier • Peier • Meier • … Find word pattern, where also the first character may be wrong! Use of the : PhoneticCommand © ai-one inc. 2012
- 8. Query chains, combinations In certain cases it may help to chain the different commands into a small workflow. Depending the project, its best KeyWordCommand to combine the ai-one commands. In the beginning on ResultSet 1 has to switch commands structures and conventional Check-Asso Check-Phonetic thinking, but then programmers are in our world very fast. ResultSet / Edit Match/Classify ResultSet 2 Use of the : combine the commands © ai-one inc. 2012
- 9. Summary: just a few commands KeyWordCommand AssoAnalyseCommant PhoneticCommand Learn & Tighten Commands Focus & others… … explain the entire semantic world! © ai-one inc. 2012
- 10. ai-one gives Better results! current linguistics and semantic solutions works only if they are feed with accurate and detailed language dependent models, and there is NO incrementally updating/learning possible! ai-one solved this challenge, ai-one’s approach works incrementally, shows the inherent (intrinsic) semantic in any language without pre programming or compelling use of ontologies and thesauri. ai-one © ai-one inc. 2011 inc. 2010
- 11. Intelligent Language Handling LWO: Dynamic and self detection ontologies Prof. Dr. habil. Ulrich Reimer, University of Applied Sciences St. Gallen, "Learning a Lightweight Ontology for Semantic Retrieval in Patient-Center Information Systems". One direct benefit and resulting application, explained also in the paper of Prof Dr. habil Ulrich Reimer is, a trend barometer that uses the ai-one core technology to observes and analyze for example the Internet (news platforms, online news, RSS feeds, blogs etc.) The trend barometer finds in context and topics discussed the current keywords, that is the semantic trends, and builds a dynamic ontology on a daily basis. Similarly, ai- one can be applied as trend barometer or analysis tool on documents or databases. This opens up the possibility to compare documents, even databases, as regards content. The number of possible applications are almost infinite. ai-one © ai-one inc. 2011 inc. 2010
- 12. Intrinsic semantic (LWO) vs.: Full-fledged ontologies [Supervised learning] - Works only with detailed models - Language dependent, - no incrementally updating Sharing / reuse of ontologies [limited possibilities] - Based on models and reservations about the quality - Language dependent - no incrementally updating Folksonomies [WEB 2.0 / semantic WEB] - No controlled quality or validation - Often incomplete or not existent, Language dependent - no incrementally updating ai-one © ai-one inc. 2011 inc. 2010
- 13. ai-one is language independent prolitterisdorerchristianvaldaandreasblick20020129seite5nummer23swissairboss12mi I0I00I0II0I0III0I0I00II0II0I00II0I0II0I000II0I0II0I00II0I00II0I0II0I000I0II0I ofür5jahreimvorauskassiertderdruckwächstcortiindeckungvonchristiandorerundandre 0II0I0I0II0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I0 asvaldazürichesistvölligunüblichdassderlohnfürfünfjahreimvorausbezahltwirdsagenre nommierteheadhunterfdppräsidentgeroldbührer53verlangtjetztschonungslostranspare 0III000I0I0III0II00I0II00II0I0I00II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000 nznochimmersagenmariocortiundseinefreundeausdemaltenverwaltungsratnocommen II00II0II0I00II0I0I0I00III000I0I0III0II00I0I0I0I0II0I00I0I0I0I0I00I0IIII0I0I0 tcortiweiltegesterninpolenundwolltenochimmernichtsagenoberaufeinenteilseines5jahr I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00II0I0I00II0I00I0I0I0 eslohnesverzichtenwirdinschweigenhüllensichauchdieehemaligenverwaltungsrätediei I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0 mmärz2001denvergoldetencortivertragaufgesetzthattenindersalärkommissionsassda malszementkönigthomasschmidheiny56herrschmidheinymöchtesichnichtäussernsola I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III ngedieuntersuchungdessachwaltersläuftlässterausrichtengleichtöntesbeimzweitenko 0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III mmissionsmitgliedgaudenzstaehelinichhabeverständnisfürdasinteresseaberesliegtan 000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0 derswissairzuinformierennichtanmirauchdieübrigenmitgliedervonvrenispoerrybislukas I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II mühlemannverkriechensichdabeiwarenauchsiebestensimbildsolcheverträgewurdenim merimplenumbesprochensagteinehemaligesvrmitglieddochcreditsuissechefmühlema 0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I nn51bequemtsichgenausowenigzueineroffeneninformationwiebankierbénédicthentsc 000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0 h53amtelefonsagterentnervtichhabedasrechtnichtszusagenwarumwollensienichtssag II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0I The CORE works on binary level enherrhentschfragensienichtweiterichgebekeinenkommentarschönentagaufwiederse henaufgehängtjetztregtsichpolitischerwiderstandwennderbetragohneauflagenüberziel III0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I eausbezahltwurdedannistdasinakzeptabelsagtfdppräsidentgeroldbührerüberfdpmitgli 0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II edcortijetztbrauchtesschnelltransparenzdenkopfschütteltauchcvppräsidentphilippstäh 0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0II elin57ichhabemühemiteinemsohohenlohndasführtzuriesigeneinkommensunterschied I0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III enimvolkdasdarfnichtseinerhaltezwarvielvonleistungslohndochdenkannmannichtimvor 000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0 ausbekommenvorauszahlungensindunüblichsagtheadhuntersandrovgianellaundfredy islervonspencerstuartichhabenochnievoneinemsolchenfallgehörtaberniemandwollted I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II enswissairjobdiesesversprechenwarwohleinlockvogeldamitcortiseinensicherennestlép 0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I ostenaufgabmariocortilukasmühlemannthomasschmidheinybénédicthentschkonzerns 000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0IIII0I0I0I0 anierermussvorgerichtzürichdervorkassevertragcortisistkeinepremiereimfallderkonkur II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I0I0I00I0I sitenbiberholdingliesssichkonzernsaniererchristianspeiserbildseinenjobmit28millionen frankengarantielohnvergoldenermusssichjetztvorgerichtverantwortenwiecortikassierte III0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II0I00I0I0I auchspeiserfrühzeitigdiegesamtesalärpauschaleerliesssichdiemillioneneinjahrvorderf 0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0I0I0I00III000I0I0III0II00I0II älligkeitauszahlendermanagerwurdeverdächtigtamzahltagbereitsvomdrohendenkonk 0I00I0I0I0I0I00I0IIII0I0I0I0II00I0II0I000II00II0II0I00II0l0l0l00l0l0l00l0l0l ursgewusstzuhabenspeisermussteseineraffgierallerdingsteuerbezahleneinewocheuh 0l0l0l0lll0l000ll0l00l0l0ll0l00l0ll0l000ll0l0l0l0l0ll000l0l0l0l0ll0l0l0l0l0l0l0l afteinestrafklagevormzürcherbezirksgerichteinezivilklagedurchdensachwalter200000f rankenmussteerzurückzahlenseinekarrierewurdejähbeendetimhe 00l0ll0l0l0l00l0l0lll0l0l0l0l0l0ll0l0l0lll0lll0l0l0l0l00l0ll0l0l0l0l00ll0l0ll0ll0l ai-one © ai-one inc. 2011 inc. 2010
- 14. ai-one plus NLP for perfect results Combine ai-one with NLP and ontology for best possible output conditioning. Categorization Find connections (NLP) (LWO) Sense, • Autonomous display of • Categorize content Meaning any kind of data based on rules Decisions • Unstructured approach • Structured approach • Recognition of all • Trained; Manually connections between updated and developed words Better decisions because ai-one! ai-one © ai-one inc. 2011 inc. 2010
- 15. A few ai-one commands solve and support: • Sentiment analyses • Social media analyses • Trend studies • Automatic classifying • Autonomic sense making • Detect unknown patterns • Answers unknown questions • Autonomic decision making .. and much more © ai-one inc. 2012
- 16. Only a few commands are need to: Explore and then Explain the world of language with basically two main commands and a few complementary commands: …that’s the ai-one LIB & API! © ai-one inc. 2012
- 17. Thank You! ai-one inc. ai-one ag ai-one gmbh 5711 La Jolla Blvd., Flughofstrasse 55, Koenigsallee 35a, Bird Rock Zürich-Kloten Grunewald La Jolla, CA 92037 8152 Glattbrugg 14193 Berlin info@ai-one.com www.ai-one.com © ai-one inc. 2012