Journals
  Publication Years
  Keywords
Search within results Open Search
Please wait a minute...
For Selected: Toggle Thumbnails
Cross-Domain Sentiment Classification Based on Representation Learning and Transfer Learning
LIAO Xiangwen, WU Xiaojing, GUI Lin, HUANG Jinhui, CHEN Guolong
Acta Scientiarum Naturalium Universitatis Pekinensis    2019, 55 (1): 37-46.   DOI: 10.13209/j.0479-8023.2018.063
Abstract1295)   HTML    PDF(pc) (1404KB)(242)       Save

Most of existing cross-domain sentiment classification methods are not expressive enough to capture rich representation of texts, and class noise accumulated during transfer process would lead to negative transfer which could adversely affect performance. To address these issues, the authors propose a method combining textual representation learning and transfer learning algorithm for cross-domain sentiment classification. This method first builds a hierarchical attention network to generate document representations with local semantic information. Afterwards, the authors utilize the class-noise estimation algorithm to detect the negative transfer samples in transferred samples and remove them. Finally, the sentiment classifier is trained on the expanded dataset from samples in target domain and transferred ones in source domain. Compared with the baselines, two experiments on large-scale product review datasets show that the proposed method is able to effectively reduce RMSE of crossdomain sentiment classification by 1.5% and 1.0% respectively.

Related Articles | Metrics | Comments0
Synthesis and Photophysical Properties of Cyano-Bridged Complexes of Ru(II)
LI Hongliang,GUO Guolin,GUI Linlin,LI Chongjia
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract597)            Save
The special property of the NO+ ligand which can be solvated by azide-assisted labilization of the nitrosyl ligand in the Ru(II) complexes was applied in the synthesis of the necomplexes: NO2-Ru(bpy)2-NC-Ru-(py)4-CN+, Cl-Ru(bpy)2-NC-Ru(py)4-CN+, NO2-Ru(bpy)2-NC-Ru(py)4-CN-Ru-(bpy)2-NO22+ and Cl-Ru(bpy)2-NC-Ru(py)4-CN-Ru(bpy)2Cl2+(bpy=2, 2'-bipyridine, py=pyridine). The UV-Visible spectra, the IR spectra and the photo-luminescence properties of these complexes were studied. It has been shown from the elemental analysis, the UV-Visible spectra and the IR spectra, that the new complexes we designed were prepared. It is very interesting that there are two different MLCT(metal-to-ligand charge transfer) process and two photo-luminescence emission bands in one molecule. These special photophysical properties were the results of the special configuration and buildups of the different blocks in the new complexes. This will present the advantage for the study of the electron transfer within those molecules.
Related Articles | Metrics | Comments0
Study on PbS/SiO2 Quantum-Dot-Glass Prepared by Sol-Gel Method
HUANG Jianbin,DAI Qinghong,GUI Linlin
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract751)            Save
The sol-gel process has been applied to preparation of small size PbS crystal doped silica glasses. Silica glasses containing 0.87%-6.6% PbS were synthesized in a process which includes the two-step hydrolysis of Si(OC2H5)4 and Pb(CH3COO) 2·3H2O, drying and heating of doped gels, and the reaction by exposing the porous glass of PbO/SiO2 to H2S gas. The sol-gel transition point and specific surface area for the doped system were measured. Surfactant microemusion is added to the Sol-gel process to obtain smaller size PbS microcrystal with narrower size distribution. From the results of transmission electron micrographs, PbS crystals were 3-6nm in diameter for methanol method and 2-3nm for microemulsion method. In the optical absorption spectra, the absorption edge exhibited a blue shift compared with that of the bulk PbS, especially for the samples made by microemusion method. Thus the quantum size effect was found for glasses containing PbS prepared by the sol-gel process.
Related Articles | Metrics | Comments0