A system for automated control of retrieval and output of music audio files, comprises a training input for receiving music audio files each having one or more associated keywords from a set of keywords. An analyser is arranged to convert keywords to M dimensional vectors in a vector space, where M is less than the total number of distinct keywords in the set of keywords. The analyser arranged to sample features of the music audio files and to produce an F dimensional vector in a vector space representing each music audio file. A machine learning module is arranged to derive a conversion between M dimensional vectors and F dimensional vectors. A sample input is arranged to receive a sample audio file, to extract features and to produce a derived F dimensional vector in vector space. A converter is arranged to convert the F dimensional vector to a derived M dimensional vector in vector space using the derived conversion. An output is arranged to allow selection and retrieval of music audio files using the derived M dimensional vector.

[1] Mathieu Barthet, David Marston, Chris Baume, György Fazekas, and Mark Sandler. Design and Evaluation of Semantic Mood Models for Music Recommendation. In Proc. International Society for Music Information Retrieval Conference, 2013. A selected set of 1760 tracks was evaluated, using one-third for testing and 2-fold cross-validation. Every pair of features were tested before taking the top 12, testing those with every other feature and repeating. The results show that 32 spectral, harmonic, rhythmic and temporal features are needed for optimum performance, but as the error converges quickly, good performance can be achieved with much fewer.

In this paper we present and evaluate two semantic music mood models relying on metadata extracted from over 180,000 production music tracks sourced from I Like Music (ILM)’s collection. We performed non-metric multidimensional scaling (MDS) analyses of mood stem dissimilarity matrices (1 to 13 dimensions) and devised five different mood tag summarisation methods to map tracks in the dimensional mood spaces. We then conducted a listening test to assess the ability of the proposed models to match tracks by mood in a recommendation task. The models were compared against a classic audio contentbased similarity model relying on Mel Frequency Cepstral Coefficients (MFCCs). The best performance (60% of correct match, on average) was yielded by coupling the fivedimensional MDS model with the term-frequency weighted tag centroid method to map tracks in the mood space.

A fibre-optic accessory with a linear drive transport system has been coupled to a near infrared (NIR) instrument to enable solid samples, in this instance increment cores from standing trees, to be scanned at 1 mm increments along the length of the sample. This allows the NIR prediction of wood properties (oven-dry chemical composition and microfibril angle) to be undertaken so that the radial profile of chemistry or microfibril angle can be determined from the pith to the bark. Calibration models provided prediction errors for microfibril angle in Pinus radiata softwood of 4.1° while for Eucalyptus globulus the error is 3.9°. The errors for prediction of chemical composition in Pinus radiata are 0.2% (arabinose) 1.1% (galactose), 2.3% (glucose), 0.7% (mannose), 0.7% (xylose) and 1.6% (lignin).

Rapidly changing market conditions and IT systems are forcing companies to adapt their business processes more dynamically than in the past. Most current commercial BPMS products lack the full capability to support changing processes. To address this challenge, we discuss three aspects of dynamic adaptation in a BPM system: model-level, instance-level, and runtime environment changes. We focus on analyzing the patterns of the instance-level navigation changes, including the preconditions, actions and consequences of each change. Based on this analysis, we propose a system design for the runtime environment to support dynamic instance-level changes. We have implemented a prototype of such a system with the motivating scenarios to demonstrate the usefulness of our proposal.

The business architecture of a service-oriented enterprise can be adequately represented through five main architectural domains: business value, structure, behavior, policy, and performance. In this paper we focus on the core business architecture, the set of essential elements in each of the five domains, and the interrelationships among these elements. The business architecture described in this paper identifies the key elements required for business reasoning and for its application to business transformation through service-oriented solutions. A business scenario involving a fictional company in the apparel business illustrates the concepts presented here.

As businesses scramble to adopt and implement Service-Oriented Architectures (SOA) it is imperative that service-oriented thinking become an integral part of the business itself. We motivate the key concepts for service-oriented business thinking through a range of examples that give a flavor of a range of businesses. In this paper we establish three key concepts - Service Agreement, Service Transaction and Service Function. These concepts are discussed in depth and are related to our ongoing work in the area of modeling business services.

Radiata pine, the predominant species in New Zealand, has a relatively low stiffness corewood zone which can make the corewood of a log unsuitable for use as structural timber. This is associated with a significant value-loss. In a mill-scale trial it was assessed whether NIR could be used to assess green timber stiffness at an early stage in the sawmill. Thus the sawing patterns could potentially be adjusted to minimize downgrading of timber. An NIR spectrometer was used to record spectra from the areas of the future boards on the cant surface, after the first two opening cuts by a headrig saw to produce a cant. The material was then tracked through the sawmill and after kiln-drying and planing the cut boards were assessed for stiffness with a three-point bending test. The stiffness values were then regressed against the respective spectra from the area of the green, uncut cant from which the board was derived using PLS modelling. The resulting correlations of r(2) = 0.54 (big logs) and r(2) = 0.57 (small logs) based on full, random cross validation, were sufficient for an economic segregation of cants into different sawing patterns based on their corewood stiffness.

Pinus radiata D. Don cants (100 or 200 mm thick × 4.8 m) from a commercial sawmill operation were assessed in the green state using near infrared (NIR) spectroscopy. Near infrared spectra were acquired along the centre line of one cant face and at 50 mm offsets to one side of the centre line. The cants were ripped to produce either 50 × 100 or 50 × 200 mm rough sawn boards, which were then kiln-dried and gauged to final dimensions. The long-span modulus of elasticity ( L MoE) on each board was determined using a four-point bending test and the corresponding NIR spectra of each board (the 50 mm edge from the cant) were regressed against the long-span MoE value using partial least squares modeling. The results are explained in terms of the potential for NIR to predict the potential upgrade to higher value products for timber recovered from the corewood zone of logs.

The business architecture of a service-oriented enterprise can be adequately represented through five main architectural domains: business value, structure, behavior, policy, and performance. In this paper we focus on the core business architecture, the set of essential elements in each of the five domains, and the interrelationships among these elements. The business architecture described in this paper identifies the key elements required for business reasoning and for its application to business transformation through service-oriented solutions. A business scenario involving a fictional company in the apparel business illustrates the concepts presented here.