Trading spaces : on the lore and limitations of latent semantic analysis

Two decades after its inception, Latent Semantic Analysis(LSA) has become part and parcel of every modern introduction to Information Retrieval. For any tool that matures so quickly, it is important to check its lore and limitations, or else stagnation will set in. We focus here on the three main aspects of LSA that are well accepted, and the gist of which can be summarized as follows: (1) that LSA recovers latent semantic factors underlying the document space, (2) that such can be accomplished through lossy compression of the document space by eliminating lexical noise, and (3) that the latter can best be achieved by Singular Value Decomposition. For each aspect we performed experiments analogous to those reported in the LSA literature and compared the evidence brought to bear in each case. On the negative side, we show that the above claims about LSA are much more limited than commonly believed. Even a simple example may show that LSA does not recover the optimal semantic factors as intended in the pedagogical example used in many LSA publications. Additionally, and remarkably deviating from LSA lore, LSA does not scale up well: the larger the document space, the more unlikely that LSA recovers an optimal set of semantic factors. On the positive side, we describe new algorithms to replace LSA (and more recent alternatives as pLSA, LDA, and kernel methods) by trading its l2 space for an l1 space, thereby guaranteeing an optimal set of semantic factors. These algorithms seem to salvage the spirit of LSA as we think it was initially conceived.

Deformable face ensemble alignment with robust grouped-L1 anchors

Many methods exist at the moment for deformable face fitting. A drawback to nearly all these approaches is that they are (i) noisy in terms of landmark positions, and (ii) the noise is biased across frames (i.e. the misalignment is toward common directions across all frames). In this paper we propose a grouped $\mathcal{L}1$-norm anchored method for simultaneously aligning an ensemble of deformable face images stemming from the same subject, given noisy heterogeneous landmark estimates. Impressive alignment performance improvement and refinement is obtained using very weak initialization as "anchors".

Data-driven impostor selection for T-norm score normalisation and the background dataset in SVM-based speaker verification

A data-driven background dataset refinement technique was recently proposed for SVM based speaker verification. This method selects a refined SVM background dataset from a set of candidate impostor examples after individually ranking examples by their relevance. This paper extends this technique to the refinement of the T-norm dataset for SVM-based speaker verification. The independent refinement of the background and T-norm datasets provides a means of investigating the sensitivity of SVM-based speaker verification performance to the selection of each of these datasets. Using refined datasets provided improvements of 13% in min. DCF and 9% in EER over the full set of impostor examples on the 2006 SRE corpus with the majority of these gains due to refinement of the T-norm dataset. Similar trends were observed for the unseen data of the NIST 2008 SRE.

The effects of L1 orthography on processing an artificial logographic script

To date, studies have focused on the acquisition of alphabetic second languages (L2s) in alphabetic first language (L1) users, demonstrating significant transfer effects. The present study examined the process from a reverse perspective, comparing logographic (Mandarin-Chinese) and alphabetic (English) L1 users in the acquisition of an artificial logographic script, in order to determine whether similar language-specific advantageous transfer effects occurred. English monolinguals, English-French bilinguals and Chinese-English bilinguals learned a small set of symbols in an artificial logographic script and were subsequently tested on their ability to process this script in regard to three main perspectives: L2 reading, L2 working memory (WM), and inner processing strategies. In terms of L2 reading, a lexical decision task on the artificial symbols revealed markedly faster response times in the Chinese-English bilinguals, indicating a logographic transfer effect suggestive of a visual processing advantage. A syntactic decision task evaluated the degree to which the new language was mastered beyond the single word level. No L1-specific transfer effects were found for artificial language strings. In order to investigate visual processing of the artificial logographs further, a series of WM experiments were conducted. Artificial logographs were recalled under concurrent auditory and visuo-spatial suppression conditions to disrupt phonological and visual processing, respectively. No L1-specific transfer effects were found, indicating no visual processing advantage of the Chinese-English bilinguals. However, a bilingual processing advantage was found indicative of a superior ability to control executive functions. In terms of L1 WM, the Chinese-English bilinguals outperformed the alphabetic L1 users when processing L1 words, indicating a language experience-specific advantage. Questionnaire data on the cognitive strategies that were deployed during the acquisition and processing of the artificial logographic script revealed that the Chinese-English bilinguals rated their inner speech as lower than the alphabetic L1 users, suggesting that they were transferring their phonological processing skill set to the acquisition and use of an artificial script. Overall, evidence was found to indicate that language learners transfer specific L1 orthographic processing skills to L2 logographic processing. Additionally, evidence was also found indicating that a bilingual history enhances cognitive performance in L2.

Broad Bean Mottle Virus: Identification, Serology, Host Range, and Occurrence on Faba Bean (Vicia Faba) in West Asia and Norm Africa

One of the faba bean viruses found in West Asia and North Africa was identified as broad bean mottle virus (BBMV) by host reactions, particle morphology and size, serology, and granular, often vesiculated cytoplasmic inclusions. Detailed research on four isolates, one each from Morocco, Tunisia, Sudan and Syria, provided new information on the virus. The isolates, though indistinguishable in ELISA or gel-diffusion tests, differed slightly in host range and symptoms. Twenty-one species (12 legumes and 9 non-legumes) out of 27 tested were systemically infected, and 14 of these by all four isolates. Infection in several species was symptomless, but major legumes such as chickpea, lentil and especially pea, suffered severely from infection. All 23 genotypes of faba bean, 2 of chickpea, 4 of lentil, 11 out of 21 of Phaseolus bean, and 16 out of 17 of pea were systemically sensitive to the virus. Twelve plant species were found to be new potential hosts and cucumber a new local-lesion test plant of the virus. BBMV particles occurred in faba bean plants in very high concentrations and seed transmission in this species (1.37%) was confirmed. An isolate from Syria was purified and two antisera were produced, one of which was used in ELISA to detect BBMV in faba bean field samples. Two hundred and three out of the 789 samples with symptoms suggestive of virus infection collected in 1985, 1986 and 1987, were found infected with BBMV: 4 out of 70 (4/70) tested samples from Egypt, 0/44 from Lebanon, 1/15 from Morocco, 46/254 from Sudan, 72/269 from Syria and 80/137 from Tunisia. This is the first report on its occurrence in Egypt, Syria and Tunisia. The virus is a potential threat to crop improvement in the region.

A novel corepressor, BCoR-L1, represses transcription through an interaction with CtBP

Corepressors play a crucial role in negative gene regulation and are defective in several diseases. BCoR is a corepressor for the BCL6 repressor protein. Here we describe and functionally characterize BCoR-L1, a homolog of BCoR. When tethered to a heterologous promoter, BCoR-L1 is capable of strong repression. Like other corepressors, BCoR-L1 associates with histone deacetylase (HDAC) activity. Specifically, BCoR-L1 coprecipitates with the Class II HDACs, HDAC4, HDAC5, and HDAC7, suggesting that they are involved in its role as a transcriptional repressor. BCoR-L1 also interacts with the CtBP corepressor through a CtBP-interacting motif in its amino terminus. Abrogation of the CtBP binding site within BCoR-L1 partially relieves BCoR-L1-mediated transcriptional repression. Furthermore, BCoR-L1 is located on the E-cadherin promoter, a known CtBP-regulated promoter, and represses the E-cadherin promoter activity in a reporter assay. The inhibition of BCoR-L1 expression by RNA-mediated interference results in derepression of E-cadherin in cells that do not normally express E-cadherin, indicating that BCoR-L1 contributes to the repression of an authentic endogenous CtBP target.

A spoofing detection method for civilian L1 GPS and the E1-B Galileo Safety of Life service

This paper describes an effective method for signal-authentication and spoofing detection for civilian GNSS receivers using the GPS L1 C/A and the Galileo E1-B Safety of Life service. The paper discusses various spoofing attack profiles and how the proposed method is able to detect these attacks. This method is relatively low-cost and can be suitable for numerous mass-market applications. This paper is the subject of a pending patent.

L1 Learning Strategy Instruction does make a difference in EFL listening: An empirical study.

In language learning, listening is the basic skill which learners should begin to develop other skills, namely speaking, reading and writing. This sequence of language learning in most English as Foreign Language (EFL) settings goes against the stream, learning first reading and writing and later listening and speaking. This study investigates the effects of cognitive, process-based approach to instructing EFL listening strategies over 11 weeks during a semester in Persian (L1). Lower intermediate female participants (N = 50) came from a couple of EFL classrooms in an English Language Institute in Iran. The experimental group (n = 25) listened to their classroom activities using a methodology that led learners through four cognitive processes (guessing, making inference, identifying topics and repetition) in Persian was basically successful in EFL listening. The same teacher taught the control group (n = 25), which listened to the same classroom listening activities without any guided attention to the learning strategy process in Persian. A pre and post listening test made by a group of experts in the language institute tracked any development in light of cognitive learning strategy instruction in EFL listening through L1. The hypothesis was that the experimental group received the guided attention in L1 during the classroom listening activities made greater gains and was verified despite the partial improvement of the control group.

Are alphabetic language-derived models of L2 reading relevant to L1 logographic background readers?

In this paper, we argue that second language (L2) reading research, which has been informed by studies involving first language (L1) alphabetic English reading, may be less relevant to L2 readers with non-alphabetic reading backgrounds, such as Chinese readers with an L1 logographic (Chinese character) learning history. We provide both neuroanatomical and behavioural evidence from Chinese language reading studies to support our claims. The paper concludes with an argument outlining the need for a universal L2 reading model which can adequately account for readers with diverse L1 orthographic language learning histories.

Enhancing foreign language learning through listening strategy delivered in L1: an experimental study

Listening used in language teaching refers to a complex process that allows us to understand spoken language. The current study, conducted in Iran with an experimental design, investigated the effectiveness of teaching listening strategies delivered in L1 (Persian) and its effect on listening comprehension in L2. Five listening strategies: Guessing, making inferences, identifying topics, repetition, and note-taking were taught over 14 weeks during a semester. Sixty lower intermediate female participants came from two EFL classrooms in an English language institute. The experimental class (n = 30) who listened to their classroom activities performed better (t value = 10.083) than the control class using a methodology that led learners through five listening strategies in Persian. The same teacher taught the students in the control class (n = 30), who listened to the same classroom listening activities without any of the above listening strategies. A pre and post listening test made by a group of experts in the language institute assessed the effect of teaching listening strategies delivered in L1. Results gathered on the post intervention listening test revealed that listening strategies delivered in L1 led to a statistically significant improvement in their discrete listening scores compared with the control group.

The impact of English as a Foreign Language (EFL) listening comprehension through first language (L1) listening strategy training

Listening is the basic and complementary skill in second language learning. The term listening is used in language teaching to refer to a complex process that allows us to understand spoken language. Listening, the most widely used language skill, is often used in conjunction with the other skills of speaking, reading and writing. Listening is not only a skill area in primary language performance (L1), but is also a critical means of acquiring a second language (L2). Listening is the channel in which we process language in real time – employing pacing, units of encoding and decoding (the 2 processes are central to interpretation and meaning making) and pausing (allows for reflection) that are unique to spoken language. Despite the wide range of areas investigated in listening strategies during training, there is a lack of research looking specifically at how effectively L1 listening strategy training may transfer to L2. To investigate the development of any such transfer patterns the instructional design and implementation of listening strategy of L1 will be critical.

The acquisition of an artificial logographic script and bilingual working memory : Evidence for L1-specific orthographic processing skills transfer in Chinese-English bilinguals

Studies of orthographic skills transfer between languages focus mostly on working memory (WM) ability in alphabetic first language (L1) speakers when learning another, often alphabetically congruent, language. We report two studies that, instead, explored the transferability of L1 orthographic processing skills in WM in logographic-L1 and alphabetic-L1 speakers. English-French bilingual and English monolingual (alphabetic-L1) speakers, and Chinese-English (logographic-L1) speakers, learned a set of artificial logographs and associated meanings (Study 1). The logographs were used in WM tasks with and without concurrent articulatory or visuo-spatial suppression. The logographic-L1 bilinguals were markedly less affected by articulatory suppression than alphabetic-L1 monolinguals (who did not differ from their bilingual peers). Bilinguals overall were less affected by spatial interference, reflecting superior phonological processing skills or, conceivably, greater executive control. A comparison of span sizes for meaningful and meaningless logographs (Study 2) replicated these findings. However, the logographic-L1 bilinguals’ spans in L1 were measurably greater than those of their alphabetic-L1 (bilingual and monolingual) peers; a finding unaccounted for by faster articulation rates or differences in general intelligence. The overall pattern of results suggests an advantage (possibly perceptual) for logographic-L1 speakers, over and above the bilingual advantage also seen elsewhere in third language (L3) acquisition.

Optimization of human papillomavirus type 16 (HPV-16) L1 expression in plants: Comparison of the suitability of different HPV-16 L1 gene variants and different cell-compartment localization

Virus-like particle-based vaccines for high-risk human papillomaviruses (HPVs) appear to have great promise; however, cell culture-derived vaccines will probably be very expensive. The optimization of expression of different codon-optimized versions of the HPV-16 L1 capsid protein gene in plants has been explored by means of transient expression from a novel suite of Agrobacterium tumefaciens binary expression vectors, which allow targeting of recombinant protein to the cytoplasm, endoplasmic reticulum (ER) or chloroplasts. A gene resynthesized to reflect human codon usage expresses better than the native gene, which expresses better than a plant-optimized gene. Moreover, chloroplast localization allows significantly higher levels of accumulation of L1 protein than does cytoplasmic localization, whilst ER retention was least successful. High levels of L1 (>17% total soluble protein) could be produced via transient expression: the protein assembled into higher-order structures visible by electron microscopy, and a concentrated extract was highly immunogenic in mice after subcutaneous injection and elicited high-titre neutralizing antibodies. Transgenic tobacco plants expressing a human codon-optimized gene linked to a chloroplast-targeting signal expressed L1 at levels up to 11% of the total soluble protein. These are the highest levels of HPV L1 expression reported for plants: these results, and the excellent immunogenicity of the product, significantly improve the prospects of making a conventional HPV vaccine by this means. © 2007 SGM.