A modified least squares algorithm to treat associations in liquid alloys

The associated model for binary systems has been modified to include volume effects and excess entropy arising from preferential interactions between the associate and the free atoms or between the free atoms. Equations for thermodynamic mixing functions have been derived. An optimization procedure using a modified conjugate gradient method has been used to evaluate the enthalpy and entropy interaction energies, the free energy of dissociation of the complex, its temperature dependance and the size of the associate. An expression for the concentration—concentration structure factor [Scc (0)] has been deduced from the modified associated solution model. The analysis has been applied to the thermodynamic mixing functions of liquid Ga-Te alloys at 1120 K, believed to contain Ga2Te3 associates. It is observed that the modified associated solution model incorporating volume effects and terms for the temperature dependance of interaction energies, describes the thermodynamic properties of Ga-Te system satisfactorily.

Volume effects and associations in liquid alloys

The regular associated solution model for binary systems has been modified by incorporating the size of the complex as an explicit variable. The thermodynamic properties of the liquid alloy and the interactions between theA ?B type of complex and the unassociated atoms in anA-B binary have been evaluated as a function of relative size of the complex using the activity coefficients at infinite dilution and activity data at one other composition in the binary. The computational procedure adopted for determining the concentration of clusters and interaction energies in the associated liquid is similar to that proposed by Lele and Rao. The analysis has been applied to the thermodynamic mixing functions of liquid Al-Ca alloys believed to contain Al2Ca associates. It is found that the size of the cluster significantly affects the interaction energies between the complex and the unassociated atoms, while the equilibrium constant and enthalpy change for the association reaction exhibit only minor variation, when the equations are fitted to experimental data. The interaction energy between unassociated free atoms remains virtually unaltered as the size of the complex is varied between extreme values. Accurate data on free energy, enthalpy, and volume of mixing at the same temperature on alloy systems with compound forming tendency would permit a rigorous test of the proposed model.

Active learning in keyword search-based data integration

The problem of scaling up data integration, such that new sources can be quickly utilized as they are discovered, remains elusive: Global schemas for integrated data are difficult to develop and expand, and schema and record matching techniques are limited by the fact that data and metadata are often under-specified and must be disambiguated by data experts. One promising approach is to avoid using a global schema, and instead to develop keyword search-based data integration-where the system lazily discovers associations enabling it to join together matches to keywords, and return ranked results. The user is expected to understand the data domain and provide feedback about answers' quality. The system generalizes such feedback to learn how to correctly integrate data. A major open challenge is that under this model, the user only sees and offers feedback on a few ``top-'' results: This result set must be carefully selected to include answers of high relevance and answers that are highly informative when feedback is given on them. Existing systems merely focus on predicting relevance, by composing the scores of various schema and record matching algorithms. In this paper, we show how to predict the uncertainty associated with a query result's score, as well as how informative feedback is on a given result. We build upon these foundations to develop an active learning approach to keyword search-based data integration, and we validate the effectiveness of our solution over real data from several very different domains.