HypothesesWhat is a Hypothesis?Characteristics of Good HypothesesSlide 4Ontology-Based Constraint Recognition of Free-Form RequestsHypothesis RefinementTable Interpretation by Sibling-Page ComparisonSlide 8Thesis Statement“Science” Thesis Statements“Engineering” Thesis StatementsHypothesesHypothesesWhat is a Hypothesis?What is a Hypothesis?A hypothesis (from Greek ὑπόθεσις) is a suggestedexplanation of a phenomenon or reasoned proposalsuggesting a possible correlation between multiplephenomena. The term derives from the ancient Greek,hypotithenai meaning "to put under" or "to suppose".The scientific method requires that one can test ascientific hypothesis. Scientists generally base suchhypotheses on previous observations or on extensionsof scientific theories.- WikipediaCharacteristics of Good Characteristics of Good HypothesesHypothesesStatement of belief, a claim, a Statement of belief, a claim, a propositionpropositionClear, precise, unambiguousClear, precise, unambiguousTestableTestableSufficient as the basis for Sufficient as the basis for experimentationexperimentationRepeatable by others who may careRepeatable by others who may careOntology-Based Constraint Ontology-Based Constraint Recognition of Free-Form RequestsRecognition of Free-Form RequestsI want to see a dermatologist between the 5th and the 10th, at 1:00 PM or after. The dermatologist should be within 5 miles from my home and must accept my IHC insurance.Appointment(x0) is with Dermatologist(x1) /\ Appointment(x0) is for Person(x2)/\ Dermatologist(x0) has Name(x3) /\ Person(x2) has Name(x4)/\ Appointment(x0) is on Date(x5) /\ DateBetween(x5, “the 5th”, “the 10th”)/\ Appointment(x0) is at Time(x6) /\ TimeAtOrAfter(x6, “1:00 PM)/\ Dermatologist(x1) is at Address(x7) /\ Person(x2) is at Address(x8)/\ DistanceLessThanOrEqual(DistanceBetweenAddresses(x7, x8), “5”) /\ Dermatologist(x1) accepts Insurance(x9) /\ InsuranceEqual(x9, “IHC”)Ontology-Based Constraint Ontology-Based Constraint Recognition of Free-Form RequestsRecognition of Free-Form RequestsOntology-based, information-extraction techniques can Ontology-based, information-extraction techniques can translate free-form requests to database queries well.translate free-form requests to database queries well.Using ontology-based, information-extraction Using ontology-based, information-extraction techniques, we can establish predicates with a recall techniques, we can establish predicates with a recall of 90% or better and arguments with a precision of of 90% or better and arguments with a precision of 95% or better.95% or better.Our ontology-based logic form generation system Our ontology-based logic form generation system preforms better than other logic form generation preforms better than other logic form generation systems, which achieve a recall within the interval systems, which achieve a recall within the interval [78%, 90%] and a precision within [81%, 87%] for [78%, 90%] and a precision within [81%, 87%] for predicate generation and achieve a recall within [65%, predicate generation and achieve a recall within [65%, 77%] and a precision within [72%, 77%] for argument 77%] and a precision within [72%, 77%] for argument generation.generation.Hypothesis RefinementHypothesis RefinementRefinement: adding qualifying statements and Refinement: adding qualifying statements and more precise measuresmore precise measuresYour initial hypothesis may be wrong—then what?Your initial hypothesis may be wrong—then what?Refining may expose that it is not worth pursuingRefining may expose that it is not worth pursuingRestrictions too complexRestrictions too complexNecessary to assume current insoluble problems must first be Necessary to assume current insoluble problems must first be solvedsolvedRefined scope and conditions may still be interesting Refined scope and conditions may still be interesting (“Much of scientific progress can be viewed as refinement (“Much of scientific progress can be viewed as refinement and development of hypotheses to fit new observations.”)and development of hypotheses to fit new observations.”)Occam’s razorOccam’s razorSimplicity in scientific theories—embrace the less Simplicity in scientific theories—embrace the less complicated hypothesiscomplicated hypothesisA hypothesis should make as few assumptions as possibleA hypothesis should make as few assumptions as possible—remove those that make no difference—remove those that make no differenceTable Interpretation byTable Interpretation bySibling-Page ComparisonSibling-Page ComparisonLabelsValuesTable Interpretation byTable Interpretation bySibling-Page ComparisonSibling-Page ComparisonSibling-Page Comparison techniques yield good Sibling-Page Comparison techniques yield good table interpretations.table interpretations.When sibling pages exist (e.g., hidden-web pages When sibling pages exist (e.g., hidden-web pages from a single site), sibling-page comparison from a single site), sibling-page comparison techniques are 90% accurate.techniques are 90% accurate.For sibling tables in the For sibling tables in the zz gold standard of ground gold standard of ground truth for web table-interpretation, sibling-table truth for web table-interpretation, sibling-table comparison techniques perform significantly comparison techniques perform significantly better (p better (p .05) than either the well known .05) than either the well known xx--technique or technique or yy-technique for table interpretation.-technique for table interpretation.Thesis StatementThesis StatementA clear and concise statement of what is to be demonstrated or developed in your thesis work. A good thesis statement makes a specific claim that your readers care about. Ideally, your introduction will give your readers the background they need to understand your thesis statement and to conclude that it matters. Estimated length is 1 to 2 sentences. Thesis statement vs. hypothesis?Science hypothesis vs. engineering proposal?(from the grad handbook)““Science” Thesis StatementsScience” Thesis StatementsA dynamic dead variable analysis will reduce the size of the hash A dynamic dead variable analysis will reduce the size of the hash table in explicit model checking by marking more variables dead table in explicit model checking by marking more variables dead compared to static dead variable analysis.
View Full Document