Sentence Compression with Supervised Machine Learning

The subject of the thesis is automatic sentence compression with machine learning, so that the compressed sentences remain both grammatical and retain their essential meaning. There are multiple possible uses for the compression of natural language sentences. In this thesis the focus is generation of television program subtitles, which often are compressed version of the original script of the program. The main part of the thesis consists of machine learning experiments for automatic sentence compression using different approaches to the problem. The machine learning methods used for this work are linear-chain conditional random fields and support vector machines. Also we take a look which automatic text analysis methods provide useful features for the task. The data used for machine learning is supplied by Lingsoft Inc. and consists of subtitles in both compressed an uncompressed form. The models are compared to a baseline system and comparisons are made both automatically and also using human evaluation, because of the potentially subjective nature of the output. The best result is achieved using a CRF - sequence classification using a rich feature set. All text analysis methods help classification and most useful method is morphological analysis. Tutkielman aihe on suomenkielisten lauseiden automaattinen tiivistäminen koneellisesti, niin että lyhennetyt lauseet säilyttävät olennaisen informaationsa ja pysyvät kieliopillisina. Luonnollisen kielen lauseiden tiivistämiselle on monta käyttötarkoitusta, mutta tässä tutkielmassa aihetta lähestytään television ohjelmien tekstittämisen kautta, johon käytännössä kuuluu alkuperäisen tekstin lyhentäminen televisioruudulle paremmin sopivaksi. Tutkielmassa kokeillaan erilaisia koneoppimismenetelmiä tekstin automaatiseen lyhentämiseen ja tarkastellaan miten hyvin erilaiset luonnollisen kielen analyysimenetelmät tuottavat informaatiota, joka auttaa näitä menetelmiä lyhentämään lauseita. Lisäksi tarkastellaan minkälainen lähestymistapa tuottaa parhaan lopputuloksen. Käytetyt koneoppimismenetelmät ovat tukivektorikone ja lineaarisen sekvenssin mallinen CRF. Koneoppimisen tukena käytetään tekstityksiä niiden eri käsittelyvaiheissa, jotka on saatu Lingsoft OY:ltä. Luotuja malleja vertaillaan Lopulta mallien lopputuloksia evaluoidaan automaattisesti ja koska teksti lopputuksena on jossain määrin subjektiivinen myös ihmisarviointiin perustuen. Vertailukohtana toimii kirjallisuudesta poimittu menetelmä. Tutkielman tuloksena paras lopputulos saadaan aikaan käyttäen CRF sekvenssi-luokittelijaa laajalla piirrejoukolla. Kaikki kokeillut teksin analyysimenetelmät auttavat luokittelussa, joista tärkeimmän panoksen antaa morfologinen analyysi.

Changes in breaded chicken and oil degradation during discontinuous frying with cottonseed oil

Discontinuous frying of breaded chicken in cottonseed oil was evaluated. Three 400 g batches of foodstuff were fried daily in a 28 L fryer at 182 °C for 4.5 minutes for 7-8 days, and the experiment was repeated three times. The total polar compounds in the oil were determined by the conventional method. Changes in the oil were determined by the quick tests Testo 265, Viscofrit and Fri-check based on physical constants, and the results were compared with those of total polar compounds obtained by the conventional method. The free fatty acids, conjugated dienes, Lovibond color, oxidative stability, fatty acid composition, and polymeric compounds were also determined. During frying, the oil samples presented 6.0-39.2% total polar compounds, 0.0-12.9% polymerized triacylglycerols, 1.3-14.5% oxidized triacylglycerols, 2.8-11.0% diacylglycerols, and 1.6-2.6% fatty acids and unsaponifiable polar compounds. The breaded chicken samples lost moisture, absorbed oil up to approximately 6%, and there were small changes in the fatty acid composition and low formation of trans-isomers. The best method for monitoring and discarding the oil was that used for the determination of total polar compounds.

Development of a bovine adenovirus type 2-based gene delivery vector /

ABSTRACT Recombinant adenoviruses are currently under intense investigation as potential gene delivery and gene expression vectors with applications in human and veterinary medicine. As part of our efforts to develop a bovine adenovirus type 2 (BAV2) based vector system, the nucleotide sequence of BAV2 was determined. Sixty-six open reading frames (ORFs) were found with the potential to encode polypeptides that were at least 50 amino acid (aa) residue long. Thirty-one of the BAV2 polypeptide sequences were found to share homology to already identified adenovirus proteins. The arrangement of the genes revealed that the BAV2 genomic organization closely resembles that of well-characterized human adenoviruses. In the course of this study, continuous propagation of BAV2 over many generations in cell culture resulted in the isolation of a BAV2 spontaneous mutant in which the E3 region was deleted. Restriction enzyme, sequencing and PCR analyses produced concordant results that precisely located the deletion and revealed that its size was exactly 1299 bp. The E3-deleted virus was plaque-purified and further propagated in cell culture. It appeared that the replication of such a virus lacking a portion of the E3 region was not affected, at least in cell culture. Attempts to rescue a recombinant BAV2 virus with the bacterial kanamycin resistance gene in the E3 region yielded a candidate as verified with extensive Southern blotting and PCR analyses. Attempts to purify the recombinant virus were not successful, suggesting that such recombinant BAV2 was helper-dependent. Ten clones containing full-length BAV2 genomes in a pWE15 cosmid vector were constructed. The infectivity of these constructs was tested by using different transfection methods. The BAV2 genomic clones did appear to be infectious only after extended incubation period. This may be due to limitations of various transfection methods tested, or biological differences between virus- and E. co//-derived BAV2 DNA.

A tour from the City of New-York, to Detroit, in the Michigan Territory, made between the 2d of May and the 22d of September, 1818 : the tour extends from New-York, by Albany, Schenectady, and Utica to Sacket's Harbor, and thence throught Lake Ontario, to St. Lawrence river, and down that stream to Hamilton village. Thence along both banks of the St. Lawrence, from Hamilton to the Thousand Islands; thence to Sacket's Harbor by water; from that place by the route of great Sodus, Geneva, Cananaigua and Batavia, to Buffalo; and from thence to Black Rock, Fort Erie, the Falls of Niagara, Queenston, Lewiston, and the memorable fields of Bridgewater and Chippewa after viewing the interesting place of Niagara, the author traversed the south shore of Lake Erie to the City of Detroit, and visited in the latter range, Dunkirk, Erie, Cleveland, Sandusky, and other places of less note. The tour is accompanied with a map upon which the route will be designated; a particular map of the Falls andRiver of Niagara, and the environs of the City of Detroit /

Errata slip inserted.

Sketches of the Battle Fields of Ontario from the War of 1812, July 15, 1876.

Accompanying caption from the Canadian Illustrated News, July 15, 1876: “We publish today a page of sketches consisting of the following battle fields in Ontario :--Lundy’s Lane where, without doubt, the hardest fought battle of 1812-15 took place, and in which more troops were engaged than in any other engagement of that war : the battle field of Stony Creek where the Canadians and Indians made a night attack on the Americans and achieved a victory over a greatly superior force and obliged the Americans to retreat back to the shelter of Old Fort George which was the scene of many engagements during the war. Beaver Dam battle field is just in the suburbs of the thriving village of Thorold, and the monument covers the remains of several soldiers whose bodies were unearthed during the building of the new Welland Canal at that place.”

Newspaper Clipping - Fields' homestead

A newspaper clipping describing a fire at the Fields' homestead near Queenston. The caption describes the house as where "Sir Isaac Brock slept the night before he was killed".

Indenture - Land sold by Hart Smith of Lincoln County to James Fields of Newark in Oxford County

Indenture regarding land sold by Hart Smith of Lincoln County to James Fields of Newark in Oxford County, First Concession, Lot no. 3. October 11, 1797.

Respuesta funcional de Toxorhynchites theobaldi (DYAR y KNAB) sobre larvas de Aedes aegypti (LINN.) vector del dengue

Tesis (Maestría en Ciencias con Especialidad en Entomología Médica) UANL

Control del vector del dengue Aedes aegypti (l) con la participación de estudiantes de primaria como educadores de salud en la cuidad de Guadalupe Nuevo Léon

Tesis (Maestria en Ciencias con Especialidad en Entomología Médica) UANL

Comparación de la eficiencia de rociados domiciliares en ultra bajo volumen (ULV) y termonebulización para el control de poblaciones adultas del vector del dengue Aedes aegypti (L.) en una colonia de Guadalupe, Nuevo León, México

Tesis (Maestría en Ciencias con Especialidad en Entomología Médica) UANL

Bionomia de Aedes aegypti L., vector del dengue en el sur del estado de Jalisco, México

Tesis (Maestría en Ciencias con Especialidad en Entomología Médica) UANL

Bionomía del vector de la malaria Anopheles vestitipennis Dyar & Knab (Díptera: Culicidae) en la cuenca del Río Polochic, Alta Verapaz, Guatemala, C.A. [por] Jaime Abraham Juárez Sandoval

Tesis (Maestría en Ciencias con Especialidad en Entomología Médica) U.A.N.L.

Evaluación en campo del depredador Mesocyclops longisetus (Thiebaud) (Copepoda: Cyclopoidea) para el control larval del vector del dengue Aedes aegypti (L.) en Monterrey, Nuevo León [por] Norma Gorrochotegui Escalante

Tesis (Maestría en Ciencias con Especialidad en Entomología Médica) U.A.N.L.

Construcción del vector adenoviral AD-UV2

Tesis (Maestría en Ciencias con Especialidad en Biología Molecular e Ingeniería Genética) UANL, 2011.